Tong WUandcoderabbitai[bot] 3ab93cd76b [Enhancement] Keep max score attention across blocks in FlashAttention for better numerical stablity (#1269) ...

* Implement max score retention across blocks in FlashAttention for improved stability

* fix manual pipeline parameters

* Update examples/flash_attention/example_gqa_fwd_varlen.py

Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>

* fix typo

* more

* fix a previous typo

---------

Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>

2025-11-17 21:53:19 +08:00

..

test_tilelang_transform_cluster_planning.py

Refactor to support upstream tvm (#595)

2025-07-30 14:00:15 +08:00

test_tilelang_transform_config_index_bitwidth.py

[Enhancement] Keep max score attention across blocks in FlashAttention for better numerical stablity (#1269)

2025-11-17 21:53:19 +08:00

test_tilelang_transform_inject_fence_proxy.py

[Language] Add Correctness and performance check scripts for V2 (#1174)

2025-11-03 02:51:59 +08:00

test_tilelang_transform_inject_set_max_nreg.py

[Refactor] Merge ThreadPartialSync and ThreadStorageSync (#741)

2025-08-23 17:07:10 +08:00

test_tilelang_transform_Inject_software_pipeline.py

[Language] Introduce StridedTensor to support non contigious torch inputs (#722)

2025-08-18 00:53:29 +08:00

test_tilelang_transform_layout_inference.py

[Language] Initial version of tilelang frontend v2 (#1120)

2025-11-03 18:15:56 +08:00

test_tilelang_transform_legalize_safe_memory_access.py

[Language] Recommend using T.dynamic instead of T.symbolic (#1076)

2025-10-20 19:48:31 +08:00

test_tilelang_transform_legalize_vectorized_loop.py

[SM100] Add sm100 GEMM layouts and tcgen05 support (#887)

2025-09-28 16:49:39 +08:00

test_tilelang_transform_let_inline.py

[TileOp] Introduce a experimental python defined T.gemm_v2 (#793)

2025-09-11 03:26:20 +08:00

test_tilelang_transform_lower_hopper_intrin.py

[Lint] Add ruff config to check for useless spaces (#807)

2025-09-13 21:06:25 +08:00

test_tilelang_transform_lower_tile_op.py

[Language] Initial version of tilelang frontend v2 (#1120)

2025-11-03 18:15:56 +08:00

test_tilelang_transform_make_packed_api.py

Refactor to support upstream tvm (#595)

2025-07-30 14:00:15 +08:00

test_tilelang_transform_multi_version_buffer.py

[Language] Initial version of tilelang frontend v2 (#1120)

2025-11-03 18:15:56 +08:00

test_tilelang_transform_pipeline_planning.py

Refactor to support upstream tvm (#595)

2025-07-30 14:00:15 +08:00

test_tilelang_transform_simplify.py

[Refactor] Deprecated T.Buffer as arguments and rename related calls into T.Tensor (#281)

2025-03-26 13:42:52 +08:00

test_tilelang_transform_thread_sync.py

[CI] allow dirty workspace for format.sh and introduce loop carry thread sync unit test (#1153)

2025-10-29 13:17:23 +08:00

test_tilelang_transform_warp_specialized.py

[Lint] Add ruff config to check for useless spaces (#807)

2025-09-13 21:06:25 +08:00