Optimizing Mixture-of-Experts (MoE) Models through Customized Torch FX Passes

Group Information

Name		Name	Last commit message	Last commit date
Latest commit History 29 Commits
log_after		log_after
log_before		log_before
.gitignore		.gitignore
README.md		README.md
all_to_all_test.py		all_to_all_test.py
aot_moe_alltoall_demo.py		aot_moe_alltoall_demo.py
aot_moe_three_layers_alltoall_demo.py		aot_moe_three_layers_alltoall_demo.py
benchmark.py		benchmark.py
compiled_autograd_demo.py		compiled_autograd_demo.py
dump_ir.py		dump_ir.py
ep_demo.py		ep_demo.py
grouped_gemm_op.py		grouped_gemm_op.py
horizontal_fusion.py		horizontal_fusion.py
main.py		main.py
model.py		model.py
moe_detection_pass.py		moe_detection_pass.py
moe_inspect_graph.py		moe_inspect_graph.py