Default Branch

e2b6b367fa · Add some fast Metal MLX SDPA kernels (#2584) · Updated 2024-11-05 16:28:00 +08:00

Branches

5ac3302fac · Prebuild all our kernels. · Updated 2024-03-18 23:39:38 +08:00

441
1

53f951f6e2 · Merge remote-tracking branch 'origin/main' into cuda-conv-tr1d · Updated 2024-03-18 04:17:56 +08:00

368
6

101a4c8389 · Moondream first bits. · Updated 2024-03-18 00:49:56 +08:00

370
1

9dc53ec8ad · Last push. · Updated 2024-03-06 06:18:30 +08:00

391
5

3f3730b657 · Preliminary implementation for the vocos model. · Updated 2024-02-15 05:16:09 +08:00

444
1

e2bf0adc2a · [WIP] Bf16 support. · Updated 2024-02-14 05:44:11 +08:00

450
1

8babfe0411 · Fixed all bugs. Improved code quality. Added tests. · Updated 2024-01-30 21:40:46 +08:00

497
6

933716b374 · Where cond get_strided_index conditionally based on function constants · Updated 2024-01-24 03:40:29 +08:00

497
1

ceaf7f1e2d · More concise macros · Updated 2024-01-23 04:20:31 +08:00

497
13

67d93b4f42 · More happy tests. · Updated 2024-01-16 01:46:18 +08:00

520
11

5637f86040 · Update yew requirement from 0.20.0 to 0.21.0 · Updated 2024-01-15 20:25:36 +08:00

520
1

cdbdb4af9c · Update yew-agent requirement from 0.2.0 to 0.3.0 · Updated 2024-01-10 22:14:03 +08:00

547
1

c2261d0222 · Merge. · Updated 2024-01-08 03:27:33 +08:00

550
4

9cd0cc1f65 · Ignore rotary for mistral. · Updated 2024-01-06 04:55:13 +08:00

557
6

289c57d600 · Removing metal fences. Increases performance substantially on m1 pro. · Updated 2023-12-29 00:31:07 +08:00

586
1

5edb07a5b1 · mps matmul · Updated 2023-12-20 09:53:18 +08:00

655
1

03641293ee · Clippy pass. · Updated 2023-12-18 22:22:43 +08:00

638
0
Included

cf27868b57 · More cleanup. · Updated 2023-12-15 08:44:22 +08:00

655
0
Included

1f23cea90c · MFA · Updated 2023-12-13 23:09:20 +08:00

668
3

a9d0657432 · Better version ? · Updated 2023-12-13 19:09:20 +08:00

663
0
Included