Default Branch

df1d3f3c9a · Fix Gemma 4 sanitize() not stripping KV projections for shared layers (#1240) · Updated 2026-05-04 22:26:18 +00:00

Branches

54449a5056 · feat(tuner): support shared 2D LoRA adapters on SwitchLinear · Updated 2026-06-01 22:10:25 +00:00    electron-rare

0
2
ui

83c408e1e6 · log only on rank 0 · Updated 2026-05-26 18:57:22 +00:00    electron-rare

0
4

a716f040b7 · Fix tokenizer test failure · Updated 2026-05-19 01:30:45 +00:00    electron-rare

0
1

ecf87bcf21 · added tqdm to save_model · Updated 2026-05-14 18:00:29 +00:00    electron-rare

0
1

92381c4eeb · log only on rank 0 · Updated 2026-05-13 11:49:39 +00:00    electron-rare

1
1

95238e5d34 · sharded gemma 4 · Updated 2026-04-23 13:53:57 +00:00    electron-rare

1
1

a1154ab94a · Fix the rope mutation in a more natural way · Updated 2026-03-18 20:34:13 +00:00    electron-rare

30
1

b37d7e7f9a · Prune models · Updated 2026-03-11 00:44:12 +00:00    electron-rare

42
1

1e0c5727b6 · add responses API with examples · Updated 2025-09-20 19:52:34 +00:00    electron-rare

255
1

72a284a4f9 · Initial example · Updated 2025-07-10 01:10:58 +00:00    electron-rare

352
1

6eb9059ce6 · Replace recursion with an explicit stack · Updated 2025-07-06 23:31:49 +00:00    electron-rare

367
5

ec3ab6bea9 · bases model can't be quantized · Updated 2025-05-07 20:28:46 +00:00    electron-rare

410
5

34940c5607 · fix test · Updated 2025-03-27 15:35:32 +00:00    electron-rare

454
4

4f4d5c80e0 · some cleanup · Updated 2025-01-09 21:11:21 +00:00    electron-rare

533
2

86d816714c · Add optional quantization types · Updated 2024-12-18 06:24:41 +00:00    electron-rare

541
1

341071a5e9 · load q4_k_m inefficiently · Updated 2024-12-04 03:54:57 +00:00    electron-rare

563
1

f1b6fd63ec · nits · Updated 2024-06-13 14:47:56 +00:00    electron-rare

645
2