Default Branch

eadd15c960 · note MAX_JOBS for flash-attn compile speed · Updated 2026-05-13 04:45:21 +00:00

Branches

dcd916b29b · bump transformers 4.57.3 · Updated 2025-12-02 15:33:44 +00:00    tocmo0nlord

247
1

08c8f3f22f · fix: total tokens and defaults in config · Updated 2025-12-02 14:38:10 +00:00    tocmo0nlord

247
2

93600fa80d · 📝 Add docstrings to feat/qwen3-vl-liger-integration · Updated 2025-11-30 18:29:28 +00:00    tocmo0nlord

249
1

83ff8bfa1a · fix: change docker miniconda install to workspace · Updated 2025-11-06 11:54:56 +00:00    tocmo0nlord

271
1

9ee7ce5c85 · set TORCH_CUDA_ARCH_LIST correctly · Updated 2025-10-29 16:59:26 +00:00    tocmo0nlord

280
2

ffb307a8a7 · update tags · Updated 2025-10-04 16:10:43 +00:00    tocmo0nlord

308
92

1d0562dedd · adding fp32 support · Updated 2025-09-26 16:32:09 +00:00    tocmo0nlord

313
1

dd85358543 · default mg · Updated 2025-09-25 20:30:23 +00:00    tocmo0nlord

314
37

3299f182ba · ungate lora with bias · Updated 2025-09-25 16:40:13 +00:00    tocmo0nlord

314
2

09725be990 · add support for CP + torch SDPA · Updated 2025-09-25 16:03:43 +00:00    tocmo0nlord

318
5

939023e661 · chunked DPO loss · Updated 2025-09-24 21:43:06 +00:00    tocmo0nlord

318
1

8564961423 · fix compile · Updated 2025-09-19 17:59:57 +00:00    tocmo0nlord

325
68

e1c7a61243 · fix reentrant when using offloading · Updated 2025-09-14 14:42:15 +00:00    tocmo0nlord

337
1

a7676af44d · hmmm · Updated 2025-09-12 17:51:10 +00:00    tocmo0nlord

330
5

e37a768960 · feat: add baseten to lmeval · Updated 2025-08-29 11:02:26 +00:00    tocmo0nlord

354
1

d3bea3a2eb · broken · Updated 2025-08-25 16:51:36 +00:00    tocmo0nlord

361
3

21ba1cd3f1 · wire up squash_position_ids · Updated 2025-08-23 20:21:28 +00:00    tocmo0nlord

362
1

78a039e1be · add depr warning for preprocess --iterable · Updated 2025-08-22 16:02:30 +00:00    tocmo0nlord

363
23
tui

c3e1882de5 · progress · Updated 2025-08-22 06:43:16 +00:00    tocmo0nlord

364
2

4870638734 · initial impl of streaming preprocessing · Updated 2025-08-19 23:10:54 +00:00    tocmo0nlord

370
3