Stanley Winata

[reposted from llvm-project for more visibility] Hey guys, I am trying to compile some llvmir for gfx908, it seems like it have a lot of spiils but is only trying...

Can we get a save method to checkpoint/save the model/save the vmfb S.T we do not need to recompile from scratch every time we run the script.

[RFC] Parameterize devices for testing

4

-Cleaner interface for device testing -Easier to test new devices

-Let's make it easier for users to browse and find models in SHARK, by putting up hyperlink in the text part of the model's name (i.e BERT, Albert, Alexnet, etc)...

Mini LM Loader Example

1

-Add example to load miniLM from SharkHUB and benchmark. -Modify TF benchmark to have growing GPU allocation. -Add shark_load helper function

Change frontend from strings to enum

3

Currently backend selection is in string, while it's great/working for now. May be confusing on what backends are valid, and may produce bugs later on (for example: typos in string...

Enhancements/Fixes to HF Benchmark Runtime

2

HF Benchmarker is a module within SHARK that enable easy testing of HF models with ONNX, Torch, TF, and SHARK-RT of course. this work is based of SharkBenchmarker for MLIR...

Add Option for setup_venv.sh to choose frontends

1

Many users have their favourite deep learning framework of choice and not use others. We should set the setup_venv.sh to be have an option to choose whether they are intending...

Missing synchronizations after Hiding semaphores from the compiler

13

### What happened? After this commit https://github.com/iree-org/iree/commit/9ef8abfee1c25e10fdc026db26c9f72662c3c4fe, our LevelZero backend seemed to break by always returning zeroes. I think we use to have similar issue until we added execution barrier...

[stream] eliding async slices

10

### Request description For our use case of updating KV-Cache during a decode step, we needed to slice a global with shape `tensor` into `tensor`. This sliced attention would be...