Speedup model init on CPU (by 10x+ for llama-3-8B as one example) #2803
self-pr-slow-ci.yml
on: pull_request
Find models to run slow tests
18s
Matrix: Run all tests for the model