pull

Add SDPA backend tests and refactor generate.py #3921

Job	Run time
gather-models-gpu	11s
test-tinystories-executorch (16-core-ubuntu)	20m 43s
compile-gguf (macos-14)	5m 24s
test-tinystories-executorch (macos-14-xlarge)	14m 48s
gather-models-cpu	11s
test-gguf-util (macos-14)	5m 43s
test-mps (macos-m1-stable)	1m 0s
test-mps-dtype (macos-m1-stable)	3m 10s
torchchat-command-load-test (macos-14)	2m 4s
runner-aoti (16-core-ubuntu)	7m 4s
runner-aoti (macos-14-xlarge)	3m 39s
runner-et (16-core-ubuntu)	9m 25s
runner-et (macos-14-xlarge)	7m 35s
test-torchao-experimental (macos-14-xlarge)	10m 42s
test-torchao-experimental-mps (macos-m1-stable)	1m 48s
test-build-runner-et-android / linux-job	15m 39s
test-cpu-compile (x86_64, stories15M)	17m 27s
test-cpu-compile (aarch64, stories15M)	11m 10s
test-cpu-aoti (x86_64, stories15M)	22m 25s
test-cpu-aoti (aarch64, stories15M)	8m 21s
test-cpu-eval-sanity-check (x86_64, stories15M)	7m 25s
test-cpu-eval-sanity-check-float16 (x86_64, stories15M)	27m 41s
test-cpu-eval-sanity-check (aarch64, stories15M)	11m 19s
test-cpu-eval-sanity-check-float32 (x86_64, stories15M)	38m 4s
test-cpu-eval-sanity-check-float16 (aarch64, stories15M)	6m 49s
test-cpu-eval-sanity-check-float32 (aarch64, stories15M)	9m 37s
test-gpu-aoti-float16 (cuda, stories15M) / linux-job	14m 9s
test-gpu-aoti-float32 (cuda, stories15M) / linux-job	14m 1s
test-gpu-aoti-bfloat16 (cuda, stories15M) / linux-job	15m 15s
test-gpu-eval-sanity-check (cuda, stories15M) / linux-job	13m 40s
test-gpu-compile (cuda, stories15M) / linux-job	23m 54s
	5h 50m 23s

Provide feedback