github/AMD-SHARK-Studio
1
1
Fork 0
mirror of https://github.com/nod-ai/AMD-SHARK-Studio.git synced 2026-02-19 11:56:43 -05:00
Code Issues Packages Projects Releases Wiki Activity
1,743 Commits 66 Branches 705 Tags
diffusers-version
Commit Graph

55 Commits

Author SHA1 Message Date
Vivek Khandelwal
ea920f2955 Add sharded Falcon support 2023-10-26 21:53:25 +05:30
Ean Garvey
e6cb5cef57 Add --additional_runtime_args option and use in OPT example. (#1855) 
* Add --additional_runtime_args option and use in OPT example.

Fix the func name. (#1838)

Co-authored-by: Sungsoon Cho <sungsoon.cho@gmail.com>
 2023-10-19 13:29:39 -05:00
Ean Garvey
caf6cc5d8f Switch most compile flows to use ireec.compile_file. (#1863) 
* Switch most compile flows to use ireec.compile_file.

* re-add input type to compile_str path.

* Check if mlir_module exists before checking if it's a path or pyobject.

* Fix some save_dir cases
 2023-10-06 23:04:43 -05:00
Abhishek Varma
cdd505e2dd [SharkInference-SharkRuntime] Adds capability to mmap vmfbs 
-- This commit is based on [VmModule.mmap() API](https://github.com/openxla/iree/pull/14124).
-- It thereby adds capability to mmap vmfbs in SHARK.

Signed-off-by: Abhishek Varma <abhishek@nod-labs.com>
 2023-06-22 20:43:40 +05:30
Ean Garvey
eb6d11cfed Change mlir dialects for tf tests to stablehlo. (#1535) 
* Change mlir dialects for tf tests to stablehlo

* Update shark_runner.py
 2023-06-14 10:43:49 -07:00
Ean Garvey
218ed78c40 Change instances of input_type='mhlo' to 'auto' (#1482) 2023-06-02 16:43:47 -05:00
Eliasj42
8111f8bf35 added ability to select gpu (#891) 
Co-authored-by: Elias Joseph <elias@nod-labs.com>
 2023-01-30 13:39:12 -08:00
Abhishek Varma
e60b4568c6 [SharkInference] Make SharkInference compile the entire module (#708) 
* [SharkInference] Make SharkInference compile the entire module

-- Previously SharkInference was compiling and providing run APIs
   for a harcoded function with function name "forward".
-- This commit makes the compiling functionality generic and now
   any function being defined within the module can be run.
-- It also creates an API to fetch all the function names defined
   within the compiled module.
-- This commit updates both web and command-line execution of Stable
   Diffusion to use new API of  SharkInference.

Signed-off-by: Abhishek Varma <abhishek@nod-labs.com>
 2023-01-03 23:25:23 +05:30
Quinn Dawkins
2bc6de650d [SD] Add support for a compiled version of the discrete Euler scheduler (#657) 
* Add Shark version of euler scheduler

* Add Shark version of euler scheduler to web ui
 2022-12-17 19:25:43 -08:00
Mehdi Amini
559928e93b Actually print the error message when SharkRunner can't initialize the driver (#482) 
Right now it would just terminate the process silently
 2022-11-13 19:08:46 -08:00
Ean Garvey
fd7baae548 Serialize torch-mlir CAPI module as bytecode instead of string. (#435) 
* Serialize torch-mlir CAPI as bytecode instead of string.

* Minor fixes to MLIR data handling in SHARK python.
 2022-10-27 14:37:15 -05:00
Ean Garvey
0bfe30d75d Fix issues with extra_args in benchmarks, pin tf==2.10 (#411) 2022-10-20 06:55:26 -07:00
Quinn Dawkins
7be1d7d0be Add option for extra arguments through SharkInference.compile (#408) 2022-10-19 15:32:48 -05:00
gpetters94
53df0620e3 Add OPT to tank (#214) 2022-10-11 11:03:56 -05:00
Ean Garvey
d82b305781 Fix issues with loading .vmfb into SharkInference 2022-09-23 09:53:13 +05:30
Prashant Kumar
4e5c592094 Enable --devices to pass through CLI. 2022-07-27 22:13:52 +05:30
Ean Garvey
6401687733 Update TF pytest marks and fix benchmark runner import. (#196) 2022-07-19 15:42:32 -05:00
Ean Garvey
c5ae01311d Make PyTorch model tests importer-independent and add benchmarks. 2022-07-15 21:54:41 -05:00
Prashant Kumar
0dcf387089 Add shark_importer for torch_models. (#183) 
All the torch_models are imported to gs::shark_tank.
Scripts have been updated.
 2022-07-12 20:38:19 -07:00
Anush Elangovan
a7435973d9 Fix black formatting 2022-06-30 20:42:02 +00:00
Prashant Kumar
83855e7b08 Capture input information from mlir_graph and generate random inputs. 2022-06-29 22:51:53 +05:30
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
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
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
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
Chi_Liu
8640c3ebfe Add shark_importer tflite module and albert_shark_test example (#110) 2022-06-09 06:17:22 -07:00
Stanley Winata
8565be9b6b ORT-HF Benchmark Integration (#101) 
-Add HF Benchmarker class.
-Add sample to benchmark HF model.

Example:
```bash
python -m benchmarks.hf_model_benchmark --num_iterations=10 --model_name="microsoft/MiniLM-L12-H384-uncased"
```
 2022-06-07 23:49:39 -07:00
Prashant Kumar
9d2ce33bd1 yapf format python files. 
Please use `yapf -i --style .style.yapf shark/*.py` to format python
files.
 2022-06-06 14:34:59 +05:30
Stanley Winata
c70ab55717 Add saving and loading of mlir to run. (#89) 2022-06-05 08:42:40 -07:00
Stanley Winata
276dcf1441 Add benchmark for TF (#87) 
-Refactor SharkBenchmarker to run TF
-Add example and test to benchmark TF
 2022-06-03 22:10:27 -07:00
yzhang93
16c50cac15 Add the option to use tuned model in shark_runner (#79) 2022-06-02 18:25:33 -07:00
Stanley Winata
c83ff66a00 Refactor get_iree_module to work with SharkBenchmark (#82) 2022-06-01 19:39:15 -07:00
Ean Garvey
8ad73d365c Add support for training TF models + fix TF BERT training example (#80) 2022-06-01 18:46:07 -07:00
Maksim Levental
f57730d2db Fix type (#75) 2022-06-01 08:30:12 -07:00
Prashant Kumar
e5517f63f5 Add the refactored training module. 2022-05-27 14:21:34 +05:30
stanley
91867e1fc2 Add Shark Benchmark 
-Introduce SharkBenchmark that bench models on regular torch, shark-py, and shark-c.
-Integrate iree-benchmark-module into Shark.
 2022-05-27 00:07:31 +00:00
Prashant Kumar
15c99adab8 Refactor the whole SharkInference API. 
1. SharkInference API is refactored to target any frontends or files,
   i.e. pytorch, tensorflow, mhlo, linalg, tosa.
2. Refactor all torch APIs.
 2022-05-26 18:55:47 +05:30
Maksim Levental
ef6f3a7b74 Eager mode with IREE (#25) 
Eager mode with IREE

reenable iree integration
 2022-05-25 21:49:19 -07:00
Prashant Kumar
737be5be09 Remove AOTModule and update vulkan iree args. 2022-05-25 21:17:17 +05:30
Prashant Kumar
6a358cb90a Update the examples removing functorch dependency. 2022-05-24 19:10:01 +05:30
Prashant Kumar
96577000a3 Remove functorch dependency. 2022-05-24 18:42:35 +05:30
Prashant Kumar
4b63e0e04a Add run_on_refbackend and gpu configs. (#52) 
`run_on_refbackend` is added to run linalg_on_tensors_backend compiled
code on torch-mlir's refbackend. Also, added gpu configs and flags.
 2022-05-23 14:18:11 -07:00
Prashant Kumar
dcc1bbac14 Add pytests-parametrize with device and dynamic option. 
Added pytests-parametrize with device and dynamic options, also
added worker flag in pytest to run tests concurrently.
 2022-05-17 11:07:56 +05:30
Vivek Khandelwal
4e7b3f66f6 Add function for models benchmarking 2022-05-10 22:37:42 +05:30
Prashant Kumar
dd4bfc9c99 Update the shark-trainer API. 
Shark-trainer API now has num_iters parameter for training.
 2022-05-10 21:25:09 +05:30
Prashant Kumar
a280b0a651 Add decompositions for bert_training. 2022-05-09 16:39:11 +05:30
Prashant Kumar
f590bd56c3 Add bert_training example. 2022-05-06 21:57:46 +05:30
Prashant Kumar
b4114c0938 Make backward API workable. 2022-05-06 14:51:44 +05:30
Prashant Kumar
08df345fbc Add make_fx whole graph extract. 2022-05-04 20:58:59 +05:30