Default Branch

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

Branches

217905c8b3 · Fix iq2_ks · Updated 2025-05-14 18:03:54 +02:00    git

4211
3698

d91316e475 · MMQ for iq6_k · Updated 2025-05-14 11:34:22 +02:00    git

4211
3698

79bdbbb3c0 · This seems to work · Updated 2025-05-13 19:01:54 +02:00    git

4211
3695

2c18ef1400 · Cleanup · Updated 2025-05-13 14:04:37 +02:00    git

4211
3690

d2362176df · Fix new CUDA FA on Touring · Updated 2025-05-12 14:01:35 +02:00    git

4211
3687

902024a64c · Fix imatrix calculation for MLA models · Updated 2025-05-12 12:20:22 +02:00    git

4211
3687

83dab6a7ce · It must be like this · Updated 2025-05-12 09:19:28 +02:00    git

4211
3688

999d991152 · Add newly created tensors to model.tensors_by_name · Updated 2025-05-11 17:03:22 +02:00    git

4211
3685

d7008ad52d · constexpr and minor changes · Updated 2025-05-11 10:21:51 +02:00    git

4211
3684

2f32589b8e · Fix race in the CUDA DeepSeek FA kernel · Updated 2025-05-11 07:03:10 +02:00    git

4211
3681

154a195f75 · Minor · Updated 2025-05-10 18:07:02 +02:00    git

4211
3682

c4e1c2c905 · CUDA: fix TG with SER · Updated 2025-05-10 10:06:48 +02:00    git

4211
3680

caf309157b · convert : adapt MiniCPM3 to separate rope_freqs insertion · Updated 2025-05-09 13:30:49 +02:00    git

4211
3680

c5ed8f4069 · Fix CUDA FlashMLA-3 with quantized KV cache · Updated 2025-05-09 08:36:38 +02:00    git

4211
3674

2565b29f33 · Handle incompatible DeepSeek GGUFs · Updated 2025-05-07 17:25:40 +02:00    git

4211
3673

93d053f7ab · Fix DeepSeek q8_0 cache · Updated 2025-05-07 11:02:05 +02:00    git

4211
3670

e6da985f02 · Fix build for Xeon Gold 6226R · Updated 2025-05-07 09:23:18 +02:00    git

4211
3669

1982beb005 · Minor tweak · Updated 2025-05-07 08:07:34 +02:00    git

4211
3674

296367a50d · Update vocab.py · Updated 2025-05-05 08:37:01 +02:00    git

4211
3672

f455ead8aa · Fix DeepSeek FA · Updated 2025-05-05 07:31:55 +02:00    git

4211
3667