Default Branch

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

ZIP  TAR.GZ

Branches

ik/sweep_bench_warmup a3975acd4c · Add batch warmup to sweep-bench · Updated 2025-05-04 10:21:19 +02:00    git	4211 3664	Compare	ZIP  TAR.GZ
ik/cuda_mmq_iq4_ks 3498ea4228 · CUDA: MMQ for iq4_ks now works · Updated 2025-05-04 08:19:23 +02:00    git	4211 3666	Compare	ZIP  TAR.GZ
ik/try_fix_367_v2 5782f1bdf0 · Yet another · Updated 2025-05-03 19:00:20 +02:00    git	4211 3663	Compare	ZIP  TAR.GZ
ik/fattn_mma 056f08182a · Use MMA for TG also when quantized · Updated 2025-05-03 14:34:56 +02:00    git	4211 3661	Compare	ZIP  TAR.GZ
ik/try_fix_367 267a12aaa0 · Trying to fix iq1_s_r4/iq1_m_r4 quantization failure · Updated 2025-05-03 12:53:39 +02:00    git	4211 3660	Compare	ZIP  TAR.GZ
s6/bitnet_name_update 0e247afcac · Fix model architecture name · Updated 2025-05-02 04:43:18 +02:00    git	4211 3658	Compare	ZIP  TAR.GZ
ik/fix_fa_avx2_bug 2b7061967a · Also this was wrong · Updated 2025-05-01 18:05:08 +02:00    git	4211 3659	Compare	ZIP  TAR.GZ
s6/qwen3_dynamic_yarn a0d10704cd · Dynamic Yarn · Updated 2025-05-01 14:29:09 +02:00    git	4211 3658	Compare	ZIP  TAR.GZ
ikawrakow-patch-1-1 6c70182744 · Updates · Updated 2025-04-30 15:10:45 +02:00    git	4211 3654	Compare	ZIP  TAR.GZ
ik/fix_358 b05c85e487 · Make it also work, not just compile · Updated 2025-04-30 10:45:07 +02:00    git	4211 3657	Compare	ZIP  TAR.GZ
ik/add_missing_enum_values_qwen3 b036119637 · Add missing enum values for qwen3 and qwen3moe · Updated 2025-04-29 10:04:38 +02:00    git	4211 3655	Compare	ZIP  TAR.GZ
ikawrakow-patch-1 1f77976476 · Update README.md · Updated 2025-04-28 16:25:48 +02:00    git	4211 3652	Compare	ZIP  TAR.GZ
ik/fattn_work_buffer 20d50172d0 · Much better FA TG with q8_0 KV cache · Updated 2025-04-28 10:26:28 +02:00    git	4211 3667	Compare	ZIP  TAR.GZ
ik/fix_div_zero 957308ca09 · Fix division by zero bug · Updated 2025-04-26 09:08:37 +02:00    git	4211 3650	Compare	ZIP  TAR.GZ
ik/fix_q41_q51_arm 78458aa83d · Fix q4_1 and q5_1 on Arm · Updated 2025-04-25 19:42:21 +02:00    git	4211 3648	Compare	ZIP  TAR.GZ
ik/cohere2 95675f6194 · Command-A needs fp32 precision for K*Q · Updated 2025-04-25 14:51:58 +02:00    git	4211 3650	Compare	ZIP  TAR.GZ
ik/arch_flags d641a5fef3 · Add ability to manually set arch flags · Updated 2025-04-25 11:41:49 +02:00    git	4211 3647	Compare	ZIP  TAR.GZ
ik/fix_arm_fa 160bf27714 · Fix FA on ARM · Updated 2025-04-25 10:58:05 +02:00    git	4211 3646	Compare	ZIP  TAR.GZ
ik/pickup_13095 f71763c2d2 · cuda: use switch in constexpr funcs · Updated 2025-04-24 17:34:00 +02:00    git	4211 3644	Compare	ZIP  TAR.GZ
ik/fix_llama4_attention 6250937c49 · Fix LLaMA-4 attention · Updated 2025-04-24 12:59:19 +02:00    git	4211 3643	Compare	ZIP  TAR.GZ