Commit Graph

1743 Commits

Author SHA1 Message Date
Prashant Kumar
b07377cbfd Refactor the shark_runner shark_inference to only support mlir_modules.
1. The shark_inference is divided into shark_importer and
   shark_inference.
2. All the tank/pytorch tests have been updated.
2022-06-28 18:46:18 +05:30
Chi_Liu
44dce561e9 SharkImporter for tflite without forward and compile (#159) 2022-06-23 22:49:35 -07:00
powderluv
4ae9331a77 Update README.md 2022-06-23 22:10:56 -07:00
Ean Garvey
0a6bc6e17f Generate test-specific repro path for each TF model test. (#158)
Set TempFileSaver path directory to shark_args.repro_dir
2022-06-23 21:58:45 -07:00
Chi_Liu
fa0aaf63c2 Fix repo_dir cannot export/write mlir file bug (#157)
-Set repro_dir as a temporary dir within current working directory.
2022-06-23 18:37:16 -07:00
Chi_Liu
b7b2dfaa8d Fix shark_importer.py iree_utils import bug (#156) 2022-06-23 16:40:11 -07:00
powderluv
a7ca9b8d68 Update parser.py 2022-06-23 15:26:17 -07:00
powderluv
2c90ddde5b Update README.md 2022-06-23 15:17:14 -07:00
Ean Garvey
71f5cfcb30 Fix package management in tank for Nightly build. (#152)
Fix formatting.
2022-06-22 15:51:57 -07:00
Chi_Liu
a635b6fbef Add tf/torch/mhlo/tosa support for SharkDownloader (#151) 2022-06-22 11:25:34 -07:00
Prashant Kumar
e8aa105b2a Divide iree_utils and do module imports on function calls. 2022-06-22 14:17:33 +05:30
Ean Garvey
08eda2ce35 Equip TF tests with save_mlir, save_vmfb, and benchmark to .csv options. (#148) 2022-06-21 21:54:44 -07:00
Chi_Liu
206ba11cc8 Add SharkDownloader for end users (#147)
* Add SharkDownloader for user

* Change tank_url to gs://shark_tank
2022-06-21 16:42:23 -07:00
powderluv
6bc9ebad24 Update test-models.yml
ignore onnx benchmarks for now
2022-06-21 13:12:59 -07:00
Prashant Kumar
334d129443 Changes to minilm to make it one on one with tensorflow. 2022-06-21 22:08:59 +05:30
Anush Elangovan
28a620f0e2 Remove ONNX importer from requirements for now.
Seems to cause install conflicts.
2022-06-21 16:15:08 +00:00
Chi_Liu
af582925f2 Enable tosa.mlir as input for SharkImporter inference (#145)
* Change shark_importer to use tosa.mlir as tflite model input from  local gen_shark_tank
2022-06-20 23:15:14 -07:00
Chi_Liu
f6e9f2d571 Generate Shark Tank for tflite (#142)
generate shark_tank for tflite
2022-06-20 14:35:35 -07:00
Anush Elangovan
c66e285e2b Enable 64bit for models 2022-06-20 20:25:29 +00:00
Ean Garvey
7198f0db0e Add TempFileSaver pytest option to tank TF tests. (#144)
add/remove pytest marks to TF tests
2022-06-20 10:00:28 -07:00
Eliasj42
aae8754ace Fixed bug where backend would segfault under specific circumstances (#134)
* updated README

* added debug printouts

* added debug printouts

* fixed bug where backend would segfault on models with multiple outputs of different sizes
fixed formating issues

Co-authored-by: Elias Joseph <elias@nod-labs.com>
2022-06-20 09:56:58 -07:00
Ean Garvey
f1edf88c33 Revert "add/remove pytest marks to TF tests"
This reverts commit b522f933a9.
2022-06-20 16:06:40 +00:00
Ean Garvey
b522f933a9 add/remove pytest marks to TF tests 2022-06-20 15:23:28 +00:00
Ean Garvey
dbabb48582 Add TempFileSaver pytest option to tank TF tests. 2022-06-20 15:02:36 +00:00
Anush Elangovan
2b3fac7519 Vulkan xfails
https://github.com/google/iree/issues/9554
2022-06-18 09:39:09 +00:00
Anush Elangovan
de223e7824 Add xfail reason 2022-06-18 08:31:26 +00:00
powderluv
fd4df79573 Update README.md 2022-06-17 14:16:39 -07:00
powderluv
0093610054 Update test-models.yml
disable a100 cpu for now
2022-06-16 16:27:11 -07:00
Ean Garvey
ce7350f5f8 Mark all TF model tests as expected to fail. (#139) 2022-06-16 16:25:18 -07:00
powderluv
2bcd261a82 updates to green / broken hearts for models 2022-06-16 14:50:56 -07:00
powderluv
918b36c177 Split perf-builder into cpu/gpu/vulkan 2022-06-16 14:08:36 -07:00
Ean Garvey
c8230fe093 Update README.md
Update testing instructions.
2022-06-16 15:49:58 -05:00
powderluv
559d9be4a5 Update README.md 2022-06-16 13:25:12 -07:00
Prashant Kumar
5fbd868296 Update readme for tf models. 2022-06-16 21:55:00 +05:30
powderluv
334b50b963 Update test-models.yml 2022-06-16 07:44:20 -07:00
Prashant Kumar
4d3acff484 Add tf masked lm models (17 models - 12 pass, 5 fail). (#135) 2022-06-16 07:28:00 -07:00
Ean Garvey
f5e67e3aa4 Update README.md 2022-06-15 14:31:57 -05:00
Prashant Kumar
b43d627b20 Add distilbert-base-uncased to tank pytorch test.
distilbert-base-uncased model validation test is added.
2022-06-15 21:14:56 +05:30
powderluv
0ddf7ad5e8 Update nightly.yml 2022-06-14 23:26:37 -07:00
Ean Garvey
dc8c221467 Add option to save .vmfb in pytest + move pytorch model tests. (#131) 2022-06-14 20:39:10 -07:00
Stanley Winata
95c2e3d6ea Enable Benchmark tests (#129) 2022-06-14 14:05:26 -07:00
powderluv
cb7007ce16 Update test-models.yml
remove pip cache hang on osx and enable benchmarks on perf-linux
2022-06-14 12:23:07 -07:00
Prashant Kumar
380b0b7c54 Add seq_classification shark module to run seq_classification
on hugging face supported models.
2022-06-14 22:36:19 +05:30
Ean Garvey
01f900a9ff Move conftest.py to fix nightly release. (#126)
* Move conftest.py into frontend directory.

* xfail albert static on vulkan backend.
2022-06-14 07:07:33 -07:00
Stanley Winata
e48642f8a5 Update benchmark command to ToM and Add to test (#125)
-Update benchmark_cl to latest benchmark_module API
-tensorflow OOM issue.
2022-06-13 23:57:05 -07:00
Ean Garvey
61b19ce091 Merge pull request #122 from nod-ai/ean-tests
Enable passing of --save-mlir cmd-line option through pytest.
2022-06-13 20:27:20 -05:00
Ean Garvey
ec13200c7c Update conftest.py 2022-06-13 19:51:21 -05:00
Ean Garvey
e872296498 Enable passing of --save-mlir cmd-line option through pytest. 2022-06-14 00:27:34 +00:00
Anush Elangovan
d81d5a35df pin numpy for onnx 2022-06-13 21:28:26 +00:00
Anush Elangovan
c6de60dab5 Disable demote 2022-06-13 16:55:28 +00:00