chenyuandGitHub beba490ba8 update mask in scaled_dot_product_attention (#8674) ...

built is_causal mask with ones_like and start with boolean, and reversed the mask -inf order

2025-01-19 15:19:23 -05:00

..

codegen

use UOp.st for kernel reduce axes (#8499)

2025-01-13 06:24:11 -05:00

engine

replace DEVICE of CONST after copy folding (#8673)

2025-01-19 11:33:39 -05:00

nn

minor cleanups in state.py [pr] (#8438)

2024-12-28 20:13:25 -05:00

renderer

tf32 tc for nv and ptx (#8635)

2025-01-17 17:43:57 -08:00

runtime

tf32 tc for nv and ptx (#8635)

2025-01-17 17:43:57 -08:00

shape

upcast overflowed idx to int64 [pr] (#8268)

2025-01-17 11:52:31 -05:00

viz

empty graph rewrite to VIZ tensor graph [pr] (#8658)

2025-01-17 11:29:33 -08:00

__init__.py

UOp.variable (#7010)

2024-10-12 18:20:44 +08:00

device.py

Use full soname for libgcc_s in CPUProgram (#8642)

2025-01-16 12:56:52 +03:00

dtype.py

typos found by gemini [pr] (#8400)

2024-12-24 22:32:25 -05:00

function.py

Tensor.mod (#8458)

2024-12-31 11:31:42 -05:00

gradient.py

Revert "Revert "move subbuffer to a rewrite rule in the scheduler (#8639)" (…" (#8643)

2025-01-16 04:30:11 -05:00

helpers.py

Disable graphs by default if using an intel macbook (#8648) (#8649)

2025-01-16 18:24:56 -08:00

multi.py

add bitcast to multi [pr] (#8652)

2025-01-17 03:17:19 -05:00

ops.py

delete unused cast/bitcast lines from ops.py [pr] (#8651)

2025-01-17 03:04:18 -05:00

py.typed

add a single py.typed (#6083)

2024-08-14 17:31:46 -07:00

tensor.py

update mask in scaled_dot_product_attention (#8674)

2025-01-19 15:19:23 -05:00