Lio Einaudi 41c4a3ed4e [Kernel][Perf] Add sm_120 (RTX PRO 6000 / RTX 50) tuned configs for batch-invariant persistent matmul (#57456) ...

Signed-off-by: LioEinaudi <zhao3024667639@gmail.com>

2026-09-19 09:30:55 -04:00

..

determinism

[Kernel][Perf] Add sm_120 (RTX PRO 6000 / RTX 50) tuned configs for batch-invariant persistent matmul (#57456)

2026-09-19 09:30:55 -04:00

hw_agnostic

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

kernels

[Quantization] Let ModelOpt MXFP8 layers load pre-processed weights Purpose (#57316)

2026-09-19 09:16:39 +00:00

layers

[ROCm][Perf] Enable the AITER GDN decode fast path for flat qkvz layouts (#53623)

2026-09-19 06:05:04 -07:00

model_loader

[Bugfix][MoE] Convert FlashInfer BF16 weights in place (#54699)

2026-09-17 22:00:04 +08:00

models

[Bugfix][Model] Fix MiMo-V2.5 fused fp8 qkv_proj sharding (pre-shard count is num_key_value_heads; MTP path too) (#57508)

2026-09-19 06:05:00 -07:00

offloader

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

warmup

[Bugfix] Isolate supplemental FlashInfer BF16 autotuning (#57285)

2026-09-17 07:38:58 +00:00

__init__.py

[Platform] Deprecate seed_everything (#31659)

2026-01-04 18:34:04 -08:00

custom_op.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

parameter.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

utils.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00