dylan e873e5f27b fix: eliminate 8x CUDA sync in MoE decode — tolist() once instead of .item() per expert

2026-08-15 13:14:11 +00:00

..

__init__.py

feat: port NaiveBatchedExperts from ds_vllm — view transpose + cublas transB

2026-08-15 13:05:45 +00:00

activation.py

feat: port NaiveBatchedExperts from ds_vllm — view transpose + cublas transB

2026-08-15 13:05:45 +00:00

naive_batched_experts.py

fix: eliminate 8x CUDA sync in MoE decode — tolist() once instead of .item() per expert

2026-08-15 13:14:11 +00:00