Daniel Garvey

Correct error tolerances for golden values on A100

4

### Issue body After #9975 error tolerances for shark on A100 have been exceeded for a few models. Here are some numbers: ``` self = , dynamic = False, device...

Failure to lower mhlo.scatter to LinalgExt

15

### What happened? may be related to #9456 but is still failing on latest release. [iree reproducer zip](https://drive.google.com/file/d/1RBhIRQTHllxS8TLUoPDtTf6m9w56Itvg/view?usp=sharing) Assuming steps are redundant with reproducers. ### Steps to reproduce your issue...

shark model testing for iree

10

Attempts to address #9537 Has the same workflow inputs as validate_and_publish_release.yml, but I wasn't sure what the intended place in the existing workflows this was meant to plug into. All...

1. because of hash checking local artifacts of the nightly build aren't being tested, the existing latest will instead, this has the downstream effect of making impossible to automatically pass...

Title says it all

Strips overloads to fix test

3

Do you think we should see if there is a strip_overloads in upstream torch and then try and move this out of heavydep? I assume its only here because of...

scalar operands type promotion test

2

Testing needed for Scalar ops where type promotion is required; cases like f32+f64.

This has a lot of changes to the infra for generating tank and running benchmarks for training workload, so please review thoroughly. As we add more support for training models,...

bump torch-mlir to e48fe4588631e7a37a2899f9d4cd5c4cbc967481

8

(allows for unifying fx_importer.py)