git/ik_llama.cpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
4,211 Commits 710 Branches 1 Tag 114 MiB

Default Branch

main

b2cb4512c5 · Create parameters overview (#1269) · Updated 2026-02-20 07:20:56 +01:00
 ZIP  TAR.GZ

Branches

ik/sync_fa

c802d59ce5 · Forgotten file · Updated 2026-01-27 16:26:42 +01:00    git

62
9
 ZIP  TAR.GZ
ik/try_split_mla

686b6f63ea · Try splitting PP MLA computation · Updated 2026-01-26 17:37:12 +01:00    git

62
1
 ZIP  TAR.GZ
ik/glm47_fa_2

04829ca412 · Adjust ncols for ADA_LOVELACE or better · Updated 2026-01-26 10:00:42 +01:00    git

64
2
 ZIP  TAR.GZ
ik/shexps_better_hybrid

109686af6f · Faster hybrid inference when shared experts · Updated 2026-01-25 15:38:54 +01:00    git

66
1
 ZIP  TAR.GZ
ik/better_fa_glm45

aff7aa0cf6 · Add condition · Updated 2026-01-25 07:52:04 +01:00    git

67
4
 ZIP  TAR.GZ
ik/rng_sampling

6e6d105d4e · Much faster rng sampling · Updated 2026-01-24 14:41:47 +01:00    git

67
1
 ZIP  TAR.GZ
ik/glm45_tg_fa_hack

c663eeaca6 · Disable when the KV cache is not f16 · Updated 2026-01-24 06:03:52 +01:00    git

69
3
 ZIP  TAR.GZ
ik/glm47_tg_fa_hack

7f5503244e · Handle quantized cache · Updated 2026-01-23 07:47:29 +01:00    git

69
2
 ZIP  TAR.GZ
ik/remove_llamafile

3a3e1638d4 · Remove llamafile remnants · Updated 2026-01-22 12:12:04 +01:00    git

70
1
 ZIP  TAR.GZ
ik/sm_graph_cuda_graphs

32f8e6a565 · Merge remote-tracking branch 'origin/main' into ik/sm_graph_cuda_graphs · Updated 2026-01-22 11:34:11 +01:00    git

72
4
 ZIP  TAR.GZ
ik/fix_batched_cublas

c37783b361 · Fix non-contiguous batched cuBLAS · Updated 2026-01-22 11:05:35 +01:00    git

76
1
 ZIP  TAR.GZ
ik/sweep_bench_nrep

a2fb4cefda · sweep_bench: set number of repetions · Updated 2026-01-21 09:33:42 +01:00    git

76
1
 ZIP  TAR.GZ
ik/correct_missing_gating_func_comments

3d5b854aee · Make comments more precise when experts gating function is missing · Updated 2026-01-21 08:08:54 +01:00    git

77
1
 ZIP  TAR.GZ
ik/correct_glm47_flash_gating_func

487411b676 · This is better · Updated 2026-01-21 06:52:10 +01:00    git

78
2
 ZIP  TAR.GZ
ik/cuda_graphs_with_overrides

a6651d017a · Change graph key · Updated 2026-01-20 16:35:53 +01:00    git

78
2
 ZIP  TAR.GZ
ik/fix_windows_no_omp

8f98961b96 · Fix build failure when OpenMP is not available · Updated 2026-01-20 12:06:25 +01:00    git

80
1
 ZIP  TAR.GZ
ik/topk_moe_fuse_bias

bc16202fc7 · Merge remote-tracking branch 'origin/main' into ik/topk_moe_fuse_bias · Updated 2026-01-20 11:47:11 +01:00    git

80
5
 ZIP  TAR.GZ
ik/glm_flash

03c0629b3c · Make FA work for mla != 0 · Updated 2026-01-20 08:58:31 +01:00    git

81
3
 ZIP  TAR.GZ
ik/sampling_refactor_sorting

1e240db2a0 · Couldn't look at it without fixing it. · Updated 2026-01-19 15:38:53 +01:00    git

82
3
 ZIP  TAR.GZ
ik/adaptive_p_2

f62e317dbe · Merge remote-tracking branch 'origin/main' into ik/adaptive_p_2 · Updated 2026-01-19 14:11:04 +01:00    git

83
8
 ZIP  TAR.GZ