deepspeedai/DeepSpeed

Pull Requests Commits
updating commit for deepspeed examples (#188)
jeffra committed 5 years ago
Verified 3ef0030a
refactoring: Deduplication (#185)
archsyscall committed 5 years ago
Verified 63da7f56
Ported BERT pre-training tutorial (#184)
Shaden Smith committed 5 years ago
Verified 5fb22a05
add representation to optimizer (#181)
kouml committed 5 years ago
Verified 3637b86b
catching up DeepSpeedExamples (#177)
Shaden Smith committed 5 years ago
Verified c1a3ec67
Support multi-output models (#170)
tjruwase committed 5 years ago
Verified 53c73fe3
Add "zero_allow_untested_optimizer" option in conf file (#173)
CalogeroZarbo committed 5 years ago
Verified 43f27332
Fix ThroughputTimer with hybrid parallelism. (#171)
Shaden Smith committed 5 years ago
Verified 20557f70
Adding static loss scaling for ZeRO. (#166)
Shaden Smith committed 5 years ago
Verified a76572dc
Export all python environment variables, not just PYTHONPATH (#165)
tjruwase committed 5 years ago
Verified 012d91df
removed restrictions for custom optimizer (#161)
CalogeroZarbo committed 5 years ago
Verified ac9cc7fe
removed session_params from deepspeed_constants.py (#162)
kouml committed 5 years ago
Verified 62d3272e
JSON configuration cleanup. (#151)
Shaden Smith committed 5 years ago
Verified 1496247a
Fix permalinks (#149)
Shaden Smith committed 5 years ago
Verified 29855c27
Web edits (#147)
Shaden Smith committed 5 years ago
Verified b84a1fa4
Web edits (#146)
Shaden Smith committed 5 years ago
Verified 4d735946
Restoring CNAME (#145)
Shaden Smith committed 5 years ago
Verified 85cc16ae
drafting Jekyll webpage (#143)
Shaden Smith committed 5 years ago
Verified 5042dc00
removing duplicated CNAME (#142)
Shaden Smith committed 5 years ago
Verified d6bc44bf
Add CNAME file. (#141)
Shaden Smith committed 5 years ago
Verified 8b12cfb4
Create CNAME
ShadenSmith committed 5 years ago
0b8d765a
PyTorch 1.3+ build support (#135)
jeffra committed 5 years ago
Verified 3d3f8d36
add skip reqs flag (#133)
jeffra committed 5 years ago
Verified e0f5cc68
Install specific apex hash (#132)
jeffra committed 5 years ago
Verified 259f894a
Enhancement: Ability to load checkpoint without loading the optimizer… (#128)
samyam committed 5 years ago
Verified 936117b5
Make lr schedulers support fp16 optimizers (#124)
tjruwase committed 5 years ago
Verified 1c0b326e
Use torch.cuda.device_count() (#126)
Shaden Smith committed 5 years ago
Verified 27d83851
Add missing link to list of schedulers (#125)
astariul committed 5 years ago
Verified 8ad8a262
Add two CLI options to help with the installation inside of conda (#113)
qwfy committed 5 years ago
Verified 5f6294bd
Fix typo (#123)
tjruwase committed 5 years ago
Verified 40885599
Terms Of Use Privacy Policy Refund Policy Impressum