olivame a2cd34dd51 1. einsum('b i d, b j d -> b i j') → torch.bmm(q, k.transpose(-1,-2)) — 直接映射 rocBLAS batched GEMM ...

2. baddbmm 把 scale 融合进 GEMM，少一次 kernel launch
3. 第二个 einsum 同理换torch.bm
每一轮加速1到两秒

2026-02-08 18:54:48 +00:00

..

base_model_inference.py

init commit

2025-09-12 21:53:41 +08:00

eval_utils.py

upload eval_utils.py file

2025-09-16 21:48:15 +08:00

profile_pipeline.py

1. einsum('b i d, b j d -> b i j') → torch.bmm(q, k.transpose(-1,-2)) — 直接映射 rocBLAS batched GEMM

2026-02-08 18:54:48 +00:00

profile_unet.py

ddim.py — torch.float16 → torch.bfloat16，修复 dtype 不匹配

2026-02-08 17:02:05 +00:00

real_eval_server.py

upload real-robot deployment code

2025-09-23 15:13:22 +08:00

world_model_interaction.py

ddim.py — torch.float16 → torch.bfloat16，修复 dtype 不匹配

2026-02-08 17:02:05 +00:00