ViperEkura 998b443aa3 refactor: dedupe fp8 meta state into per-operand rings ...

- collapse FP8TensorMeta's 12 slots + 6 copy-paste methods into three _ScaleRing objects (hist/idx/scale/initialized + update/seed)
- skip meta allocation entirely on the DynamicScaling path (zero rings, scales measured inline)
- drop write-only FP8State._last_device and unused E4M3_MAX alias

2026-08-24 21:19:45 +08:00

..

backend

refactor: rework attention backend resolution

2026-08-23 14:47:02 +08:00

lib

refactor: move compiled kernel .so files into extension/lib

2026-07-31 15:36:32 +08:00

ops

feat: static fp8 weights and bias with fused epilogue

2026-08-24 19:25:23 +08:00

__init__.py

refactor: separate extension ops and backends

2026-08-16 21:15:52 +08:00

fp8.py

refactor: dedupe fp8 meta state into per-operand rings

2026-08-24 21:19:45 +08:00

loader.py

refactor: unify kernel module loading and packaging

2026-08-23 15:57:04 +08:00