microsoft/DeepSpeed

Pull Requests Commits

jeffra committed 4 years ago

013df6a9

force amd install via sudo (#1815)

jeffra committed 4 years ago

Verified a32e9b33

Invoke hipify from op builder for JIT extension builds (#1807)

jithunnair-amd committed 4 years ago

Verified 350d74ca

add GitHub URL for PyPi (#1812)

andriyor committed 4 years ago

Verified d7684f4e

Refactor moe/non-moe gradient reduction (#1811)

tjruwase committed 4 years ago

Verified 9f7126fc

Synchronize the GPUs for the text-generation inference test (#1805)

RezaYazdaniAminabadi committed 4 years ago

Verified 60fc06c6

AMD support (#1430)

jeffra committed 4 years ago

Verified c3c8d5dd

Minjiaz/mos doc (#1802)

awan-10 committed 4 years ago

Verified f0304bd1

bump DSE to latest commit

jeffra committed 4 years ago

3401d251

Refactor MoE and Groups API to simplify model creation and mangement (#1798)

awan-10 committed 4 years ago

Verified c0af6d90

Bump nokogiri from 1.12.5 to 1.13.3 in /docs (#1794)

dependabot[bot] committed 4 years ago

Verified a254f39e

Generalize the model input format of the flops profiler get_model_profile() API (#1768)

cli99 committed 4 years ago

Verified 2151c787

bf16 is supported w. zero 1, fix assert (#1779)

jeffra committed 4 years ago

Verified 8eef742f

Fix CPU-Offload: Send groups of parameter lists as the FP16 parameters (#1774)

RezaYazdaniAminabadi committed 4 years ago

Verified 5ca26277

[save_16bit_model] add missing prologue (#1741)

stas00 committed 4 years ago

Verified baef92e2

Prepare zero-3 optimizer for ckpt load/save (#1750)

tjruwase committed 4 years ago

Verified 5fe5b38e

unset torch arch list for JIT mode (#1765)

jeffra committed 4 years ago

Verified 674c7588

[Doc] Add async I/O op (#1754)

tjruwase committed 4 years ago

Verified 4f96ffd9

Load MoE checkpint at deepspeed inference-engine (#1759)

RezaYazdaniAminabadi committed 4 years ago

Verified 841f99d1

Fix autotuning config (#1660)

cli99 committed 4 years ago

Verified cccc5450

fixing the inference build path when pre-building the inference op (#1755)

RezaYazdaniAminabadi committed 4 years ago

Verified d3cad051

fix pytest issue (#1764)

jeffra committed 4 years ago

Verified 56eac382

fixing a bf16 support issue (#1760)

duli2012 committed 4 years ago

Verified 97f8a9eb

Loosen requirement on packaging dependency (#1758)

jeffra committed 4 years ago

Verified dbe8ee16

Improve how runner parses env var file (#1747)

liamcli committed 4 years ago

Verified dac9056e

Move param_shapes to model files (#1732)

tjruwase committed 4 years ago

Verified 135a6256

separate add and mul flops compute function (#1745)

cli99 committed 4 years ago

Verified ba9c4cc7

Fix the tensor-slicing with multi-GPU inference and kernel-injection (#1724)

RezaYazdaniAminabadi committed 4 years ago

Verified fa073576

[debug log] disable (#1736)

stas00 committed 4 years ago

Verified bea701a1

[distributed] print dist init only on rank 0 (#1737)

stas00 committed 4 years ago

Verified e26de20b

Newer Older