Default Branch

41abbfd599 · qwen4exp: enable rms_norm + mul fusion (#28896) · Updated 2026-09-14 16:16:30 +02:00

Branches

b8b8738f92 · Merge branch 'master' into xsn/fix_longest_edge_with_limit · Updated 2026-08-07 15:38:55 +02:00

678
3

fbb29b3f98 · address review comments · Updated 2026-08-07 14:40:02 +02:00

656
5

1621aade61 · server: (router) do not evict busy models · Updated 2026-08-07 14:38:06 +02:00

657
1

1a0e158115 · fix it · Updated 2026-08-06 13:30:01 +02:00

678
2

6751477adf · cont : avoid copy_state · Updated 2026-08-06 09:19:52 +02:00

680
14

f8fa4df82a · test: add duplicated index case for ggml_mul_mat_id · Updated 2026-08-06 01:16:17 +02:00

678
1

f7655b451d · server: fix empty response for /cors-proxy · Updated 2026-08-06 01:09:24 +02:00

678
1

91f82eb1b4 · mtmd: fix granite 4v grid assembly · Updated 2026-08-06 00:42:06 +02:00

679
1

0377fc17ad · sampler : add llama_sampler_copy · Updated 2026-08-05 13:40:53 +02:00

692
2

e77293d4cc · mtmd: weave deepseek-ocr rows in one shot instead of per row (#26615) · Updated 2026-08-05 13:07:53 +02:00

828
2

8045779cff · cleanup · Updated 2026-08-05 10:22:33 +02:00

691
19

ac6c66699d · vendor : apply patches for subprocess.h · Updated 2026-08-05 01:34:32 +02:00

692
1

58137f0f47 · quant : allow quantization of ffn_gate_inp tensors · Updated 2026-08-04 17:11:42 +02:00

700
1

285941de20 · nits 2 · Updated 2026-08-04 16:49:43 +02:00

699
3

56fe6a8187 · models : fix dflash wo_a reshape on load · Updated 2026-08-04 15:41:42 +02:00

700
1

201e50cc20 · clean up comments · Updated 2026-08-04 15:18:19 +02:00

716
56

eb75c420f0 · Shared default of 64 for history-based samplers, remove context_size · Updated 2026-08-04 08:02:53 +02:00

710
2

3dfb7c7fad · server: simplify get_info result handling · Updated 2026-08-03 18:22:56 +02:00

721
4

7f5903206c · correct to 9931 · Updated 2026-08-03 12:18:50 +02:00

729
3

ddd9cdb158 · reuse str_to_lower from server-models · Updated 2026-08-02 20:41:31 +02:00

745
2