Eric BuehlerandGitHub c964e622d9 Add MXFP4 gather gemm support (#1615) ...

* Add mxp4 metal quantized matmuls

* Add scale to float

* Update quantized kernel

* Updated

* Fix compilation error

* Add mxfp4 quant method

* Update check

* Fix deser

2025-08-04 12:39:40 -04:00

..

bitsandbytes

DSV3/R1 fixes (#1173)

2025-03-08 16:56:27 -05:00

blockwise_fp8

Reworked and improved attention chunking (#1591)

2025-08-03 07:52:33 -04:00

gptq

Add the Qwen 3 and Qwen 3 MoE models! (#1285)

2025-04-29 06:23:50 -04:00

hqq

Reworked and improved attention chunking (#1591)

2025-08-03 07:52:33 -04:00

marlin

Reworked and improved attention chunking (#1591)

2025-08-03 07:52:33 -04:00

ops

Unify bitwise operations (#1288)

2025-04-25 16:52:03 -04:00

rotary

Format kernels (#1251)

2025-03-28 21:07:11 -04:00

scalar_fp8

Reworked and improved attention chunking (#1591)

2025-08-03 07:52:33 -04:00

vector_fp8

Add MXFP4 gather gemm support (#1615)

2025-08-04 12:39:40 -04:00