microsoft/DeepSpeed

Pull Requests Commits

Rename nv-torch-latest-cpu workflow to cpu-torch-latest

loadams committed 1 year ago

fa4829fc

Add fp16 support of Qwen1.5 models (0.5B to 72B) to DeepSpeed-FastGen (#5219)

ZonePG committed 1 year ago

Verified bcc617a0

[zero++] Synchronize at the end of secondary partitioning and simplify the logic (#5216)

ByronHsu committed 1 year ago

Verified 4578c249

Allow specifying MII branch on MII CI (#5208)

mrwyattii committed 1 year ago

Verified 857584fd

Fix assertion to run pipeline engine with a compiled module (#5197)

tohtana committed 1 year ago

Verified aed599b4

Update version.txt after 0.13.4 release (#5196)

mrwyattii committed 1 year ago

Verified 07a1ede4

Add script to check for `--extra-index-url` (#5184)

loadams committed 1 year ago

Verified 5115df3f

[CPUAdam] Update full_precision_optimizer_states in docstring (#5181)

rohan-varma committed 1 year ago

Verified 8c42a302

Disable ninja by default (#5194)

mrwyattii committed 1 year ago

Verified 48749411

allow debug/experimental compiler backends (#5191)

tohtana committed 1 year ago

Verified d274ebf3

Fixes for `--extra-index-url` (#5183)

loadams committed 1 year ago

Verified 5d754606

Update version.txt after v0.13.3 release (#5185)

loadams committed 1 year ago

Verified daa5a05a

Switch cpu-inference workflow from --extra-index-url to --index-url (#5182)

loadams committed 1 year ago

Verified afdf028a

Reduce ds_id name length (#5176)

jomayeri committed 1 year ago

Verified 4ec8762b

Inference V2 Human Eval (#4804)

lekurile committed 1 year ago

Verified 2a6c58df

Fix typos in blogs/ (#5172)

jinyouzhi committed 1 year ago

Verified 8d0150d9

Write multiple items to output file at once, in distributed data analyzer. (#5169)

bm-synth committed 1 year ago

Verified d5fa87ff

Fix allreduce for BF16 and ZeRO0 (#5170)

tohtana committed 1 year ago

Verified dd3690c5

Update flops profiler to handle attn and __matmul__ (#4724)

KimmiShi committed 1 year ago

Verified 98c96e79

Use ninja to speed up build (#5088)

jinzhen-lin committed 1 year ago

Verified b00533e4

Fix gradient clipping (#5150)

tohtana committed 1 year ago

Verified 005afe12

MOE: Fix save checkpoint when TP > 1 (#5157)

mosheisland committed 1 year ago

Verified a84d07c5

DeepSpeedZeroOptimizer_Stage3: remove cuda specific optimizer (#5138)

nelyahu committed 1 year ago

Verified 7f0950f8

Distributed in-memory map-reduce for data analyzer (#5129)

bm-synth committed 1 year ago

Verified e977c7d6

get_grad_norm_direct: fix a case of empty norm group (#5148)

nelyahu committed 1 year ago

Verified f062a1b2

Pin to PyTest 8.0.0 (#5163)

loadams committed 1 year ago

Verified fa3662f7

Update deprecated HuggingFace function (#5144)

mrwyattii committed 1 year ago

Verified a37e59b5

Fix UserWarning: The torch.cuda.*DtypeTensor constructors are no long… (#5018)

ShukantPal committed 1 year ago

Verified 177dc143

TestEmptyParameterGroup: replace fusedAdam with torch.optim.AdamW (#5139)

nelyahu committed 1 year ago

Verified a3be0d44

Added missing `isinstance` check in PR 5112 (#5142)

bm-synth committed 1 year ago

Verified 9bd62e0b

Older