Prashant Kumar
f9864b4ce1
Modify the directory structure to remove conflicts.
...
The directory structure is modified, also disabled the aot_module
unless the shark_trainer is deployed.
2022-04-04 21:23:33 +05:30
Prashant Kumar
59485f571e
Update readme.md of shark_runner
...
Updated the readme.md of shark_runner and added setup.py for pip
installation.
2022-04-01 01:00:32 +05:30
Prashant Kumar
d11accd6ad
Update model tracking.
2022-03-30 19:46:22 +05:30
Ramiro Leal-Cavazos
7d429e3c92
Update torch-mlir utils + remove unecessary imports in shark_runner
...
This commit updates the `torch_mlir_utils.py` to use the new
torch-mlir python package structure. This commit also removes
unecessary imports from the Python scripts in the shark_runner directory.
2022-03-30 19:42:19 +05:30
Prashant Kumar
d5a626a4c6
Changing README.md.
2022-03-30 19:37:31 +05:30
Prashant Kumar
f30d24ab2b
Update README.md with model tracking information
...
Hugging face models and TorchVision models tracking sheet has been added.
2022-03-30 19:33:38 +05:30
Prashant Kumar
1cdc9cd198
Add bert aot forward example.
...
Add option to attach custom inference function to aot_module.
Also, added an example of bert inference to run via aot.
2022-03-29 21:41:18 +05:30
Eliasj42
29113c4b71
Merge pull request #9 from NodLabs/rest-api-rework
...
fixed segfaults when running iree
2022-03-25 14:30:17 -07:00
Prashant Kumar
77a2525ace
Fix bug related to aot module.
...
Single element is returned as is and tuple is not created for that.
2022-03-25 18:32:10 +05:30
Elias
6a2efa35a5
fixed segfaults when running iree
2022-03-23 20:59:23 +00:00
Eliasj42
0e762fabbd
Merge pull request #8 from NodLabs/rest-api-rework
...
reworked how the server and client were structured
2022-03-22 19:47:42 -07:00
Elias
d447d874dd
reworked how the server and client were structured so that they could be more usable
2022-03-22 21:42:36 +00:00
powderluv
cf2e1bc4f0
Merge pull request #7 from pashu123/mulinps
...
Add support to pass multiple inputs.
2022-03-22 10:10:50 -07:00
Prashant Kumar
f3f0c36902
Add support to pass multiple inputs.
2022-03-22 16:00:01 +00:00
Elias
cc9eea9c1f
added functionality to run from c rather than the command line
2022-03-19 00:10:26 +00:00
Prashant Kumar
a89cf2d65f
Add examples of shark_inference via torch.script, torch.jit_trace, and
...
aot.
2022-03-15 21:04:26 +05:30
Elias
2416936ecd
added parallel cuda execution
2022-03-15 00:01:57 +00:00
Prashant Kumar
e6115da192
Add support for training in shark_runner.
...
Added support for training via AOT_module in shark_runner.
2022-03-13 21:50:46 +05:30
Prashant Kumar
fba169f456
Add support of AOT_Module in shark_runner.
...
Added support of AOT_Module for inference on simple example.
2022-03-12 22:07:30 +05:30
Prashant Kumar
7a0296f359
Added missing set_dep_pypaths scripts.
2022-03-11 22:21:41 +05:30
Prashant Kumar
9f252d688a
[S-RUNNER] Initial commit of Shark Runner Inferene API.
...
Added shark-runner inference API, fully functional given only one input.
2022-03-11 22:21:41 +05:30
Elias
e5174ca097
wip for dynamic indexes
2022-03-10 18:45:09 +00:00
Elias
5223ff69fe
removed cmake artifacts (somehow initial commit on this didn't go through)
2022-03-09 19:37:35 +00:00
Elias
3ea6174583
now runs on cuda
2022-03-09 19:33:29 +00:00
elias
d0fd43b897
Initial Commit
2022-03-04 16:51:50 -08:00
