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

36682005e4 · grad reduction in fp32 · 最后更新于 2022-02-23 00:19:39 +00:00    frostbyte_neo

2532
3

fb0dc00f88 · bf16 updates · 最后更新于 2022-02-18 08:21:09 +00:00    frostbyte_neo

2532
1

0c9b949c32 · fix quantizer to work with HF checkpoints · 最后更新于 2022-02-15 17:27:56 +00:00    frostbyte_neo

2544
1

d9ddaeb2ab · Update sparse_attn.py · 最后更新于 2022-02-08 01:24:07 +00:00    frostbyte_neo

2545
1

51a12669a3 · Remove debug prints · 最后更新于 2022-02-02 09:30:08 +00:00    frostbyte_neo

2717
15

71053f96aa · add some gemm kernel support for bfp16 · 最后更新于 2022-01-29 02:03:03 +00:00    frostbyte_neo

2556
1

3fde42481a · add sparse attn as extra · 最后更新于 2022-01-27 07:26:20 +00:00    frostbyte_neo

2553
2

db207d872e · Merge branch 'master' into olruwase/restore_from_bit16_weights · 最后更新于 2022-01-21 02:15:01 +00:00    frostbyte_neo

2558
2

24fe7002eb · unit test, remove exception, add notes · 最后更新于 2022-01-21 01:02:07 +00:00    frostbyte_neo

2559
1

8639b3c476 · preserve cuda visible devices order · 最后更新于 2022-01-20 01:39:06 +00:00    frostbyte_neo

2560
1

b4d895f990 · Fix inference api & add more description on inference engine tutorial · 最后更新于 2022-01-19 22:55:52 +00:00    frostbyte_neo

2561
1

6760338410 · minor update · 最后更新于 2022-01-19 06:02:29 +00:00    frostbyte_neo

2564
6

7b24585fdb · Merge branch 'master' into pr_moe_tutorial · 最后更新于 2022-01-19 01:18:09 +00:00    frostbyte_neo

2565
5

9878c95836 · edits · 最后更新于 2022-01-19 01:16:39 +00:00    frostbyte_neo

2566
4

bd0be5a294 · Update moe-inference-tutorial.md · 最后更新于 2022-01-19 00:34:31 +00:00    frostbyte_neo

2566
2
staging-moe-next-v1
Formatting / formatting (push) Has been cancelled
Build / nv-torch12-p40 (push) Has been cancelled
Build / nv-torch18-v100 (push) Has been cancelled
Build / nv-transformers-v100 (push) Has been cancelled

dab741a52f · simplify moe ckpt unit test · 最后更新于 2022-01-18 23:28:47 +00:00    frostbyte_neo

2567
7

aceaa5b3e4 · fix the inference injection policies & add some description on the inference-engine parameters · 最后更新于 2022-01-10 19:56:53 +00:00    frostbyte_neo

2571
1

dbd08236a6 · formatting · 最后更新于 2022-01-05 22:25:51 +00:00    frostbyte_neo

2580
2

7a459056fd · fix the top1gating logic when use_rts is set to false · 最后更新于 2021-12-22 18:58:57 +00:00    frostbyte_neo

2587
1

c160e9d403 · Merge branch 'master' into jeffra/z3-new-param · 最后更新于 2021-12-08 15:26:14 +00:00    frostbyte_neo

2604
4