Default Branch

8e53fcefd2 · webgpu : avoid crash when offset is not multiple of 4 in WebGPU ggml_backend_tensor_get() implementation (#28045) · Updated 2026-08-31 16:04:38 +02:00

Branches

d21c273fcb · Fix python type-checker · Updated 2026-08-31 15:31:00 +02:00

1
2

58df04c454 · metal : add top-k radix implementation · Updated 2026-08-31 14:50:20 +02:00

9
1

d18d296004 · ui : mark llama-app-reusable code with a LLAMA-APP-REUSE tag · Updated 2026-08-31 14:47:05 +02:00

2
14

20e251830f · ui : add model download pipeline · Updated 2026-08-31 14:47:05 +02:00

2
13

9b5a2ba447 · ui : move catalog url to models-discover constants · Updated 2026-08-31 14:47:05 +02:00

2
10

3259c9c3ce · ui : add model compatibility estimation · Updated 2026-08-31 14:47:05 +02:00

2
11

f286815f78 · ui : lowercase reasoning capability and use sidecar enums in tests · Updated 2026-08-31 14:47:05 +02:00

2
6

c1a7930d5e · ui : add format:files script for formatting changed files only · Updated 2026-08-31 14:47:05 +02:00

2
3

9d0357807e · fix fp4 extension use switch being overwritten by generic shader · Updated 2026-08-31 10:36:34 +02:00

20
8

28bd87b2cd · rpc: allow -sm tensor · Updated 2026-08-30 18:02:46 +02:00

32
1

7f33a01dd0 · autoscale n_ctx_train when yarn scaling specified · Updated 2026-08-30 16:37:15 +02:00

17
1

049c5f7c03 · metal : add fa-vec tuning for M2 Max · Updated 2026-08-30 12:11:52 +02:00

20
1

965e57103f · fix shmem support function, clean up comments · Updated 2026-08-29 14:05:48 +02:00

37
50

cda6f9e3a0 · Merge remote-tracking branch 'origin/master' into xsn/tensor_read_lazy_improve · Updated 2026-08-28 22:59:01 +02:00

42
3

f91123d2d0 · qwen4exp: fix sparse-attention block selection · Updated 2026-08-28 20:38:14 +02:00

48
1

248ae7ee4a · ci : generalize graphics device name pattern · Updated 2026-08-28 16:44:42 +02:00

47
4

18f3082e88 · metal : rename variable · Updated 2026-08-28 16:36:14 +02:00

47
2

5851394a75 · cont : ws · Updated 2026-08-28 14:54:13 +02:00

48
10

6bfbc89820 · convert: prevent ndarray conversion in LazyChunkedTensor · Updated 2026-08-28 11:25:42 +02:00

51
1

a22c1d344e · ui : apply ESLint formatting to models discover components · Updated 2026-08-28 09:42:04 +02:00

55
17