分支列表

默认分支

71d316d608 · Deprecate unused DeepSpeed features (#8490) · 最后更新于 2026-09-11 22:50:44 +00:00

分支列表

f832089919 · ]Merge branch 'master' of github.com:microsoft/DeepSpeed into olruwase/trainable_parameters · 最后更新于 2021-08-23 16:07:16 +00:00    frostbyte_neo

2752
2
elasticity-v2
Build / build (push) Has been cancelled

c19651267f · formatting · 最后更新于 2021-08-17 18:59:16 +00:00    frostbyte_neo

2756
2

cb452ccf18 · Merge branch 'master' into olruwase/align_rrg_rs_param_order · 最后更新于 2021-08-05 19:32:31 +00:00    frostbyte_neo

2764
3

92d27e3c1e · Merge branch 'olruwase/round_robin_gradient_option' of github.com:microsoft/DeepSpeed into olruwase/round_robin_gradient_option · 最后更新于 2021-07-28 21:40:20 +00:00    frostbyte_neo

2770
6
jeffra/fs-diverge
Build / build (push) Has been cancelled

6ae756c03f · ignore overlap/contiguous_gradients if using zero 1 (#1246) · 最后更新于 2021-07-27 00:13:26 +00:00    frostbyte_neo

2772
0
已包含
tmp
Build / build (push) Has been cancelled

6ae756c03f · ignore overlap/contiguous_gradients if using zero 1 (#1246) · 最后更新于 2021-07-27 00:13:26 +00:00    frostbyte_neo

2772
0
已包含
jeffra/engine-xthru
Build / build (push) Has been cancelled

8e4dd794a0 · formatting · 最后更新于 2021-07-23 18:55:55 +00:00    frostbyte_neo

2793
9

79ee82f6ed · Merge branch 'master' into reyazda/fix-inference-api · 最后更新于 2021-07-20 20:03:13 +00:00    frostbyte_neo

2776
6
samyamr/gpt3-finetuning-mixed-precision
Build / build (push) Has been cancelled

f03d605391 · Detects tensor outputs hidden inside dictionary returned by modules and registers Z3 hooks correctly. Gives the ability to assign different model facing data types for Z3 parameters during Init. The dtype of partitioned parameters is still controlled by DS config · 最后更新于 2021-07-19 17:26:21 +00:00    frostbyte_neo

2843
10
olruwase/gpt3-finetuning
Build / build (push) Has been cancelled

7e5b51e61e · Minor tweak · 最后更新于 2021-07-19 17:24:55 +00:00    frostbyte_neo

2843
12
megatron2.4-3d
Build / build (push) Has been cancelled

d26c258b08 · added shaden's set_train_batch_size patches, plus formatting · 最后更新于 2021-07-13 17:38:48 +00:00    frostbyte_neo

2796
12

7030346c77 · Merge branch 'master' into olruwase/zero2_offload_rrb_divergence · 最后更新于 2021-07-13 00:19:57 +00:00    frostbyte_neo

2783
12
samyamr/gpt3-finetuning
Build / build (push) Has been cancelled

302eb67c5c · GPT-3 finetuning seems to be working · 最后更新于 2021-07-06 21:06:14 +00:00    frostbyte_neo

2843
9
reyazda/triton-new-sparse
Build / build (push) Has been cancelled

af6c640917 · minor fixes · 最后更新于 2021-07-05 23:19:12 +00:00    frostbyte_neo

2811
3
reyazda/test-transformer
Build / build (push) Has been cancelled

fa29f783a4 · add more tests for the transformer kernel · 最后更新于 2021-06-29 16:09:54 +00:00    frostbyte_neo

2805
1
transformer/large-seq-support
Build / build (push) Has been cancelled

a0bd66c605 · support higher sequence length in softmax backward (>2048 & <8192) · 最后更新于 2021-06-18 20:33:45 +00:00    frostbyte_neo

2804
1

bed599bc7d · Merge branch 'master' into amawa/zero-inf-refactor · 最后更新于 2021-06-03 12:09:17 +00:00    frostbyte_neo

2824
5

e7378a462e · Merge branch 'master' into sparse-attn/support-latest-triton · 最后更新于 2021-06-02 18:48:43 +00:00    frostbyte_neo

2827
4
olruwase/grad_accum_loss
Build / build (push) Has been cancelled

ca1bccd9e1 · Don't expose loss scaling to clients · 最后更新于 2021-05-29 12:36:25 +00:00    frostbyte_neo

2827
1

50cce84a1e · Merge branch 'olruwase/assert_unused_parameters' of github.com:microsoft/DeepSpeed into olruwase/assert_unused_parameters · 最后更新于 2021-05-07 16:12:17 +00:00    frostbyte_neo

2854
8