Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
914 commits
Select commit Hold shift + click to select a range
c199eda
refactor to use mem_access (#2317)
mrwyattii Sep 14, 2022
95d1151
add quant unit test (#2315)
GuanhuaWang Sep 14, 2022
cf638be
only override forward if using cuda-graph (#2291)
jeffra Sep 14, 2022
1592381
Add more options to inference benchmark (#2325)
mrwyattii Sep 16, 2022
0f0a7a5
bump to 0.7.4
jeffra Sep 19, 2022
80b10d0
MOE residual matmult unit test (#2323)
samadejacobs Sep 19, 2022
12e1cb8
MOE matmult with memaccess (#2336)
samadejacobs Sep 21, 2022
48c5220
Refactor residual add kernels (#2333)
arashb Sep 21, 2022
954e0c6
mem access for quantize kernel (#2331)
GuanhuaWang Sep 22, 2022
b76e0f4
increase min pre-commit versions (#2346)
jeffra Sep 22, 2022
3d097bb
Extend scratch buffer for long prompts (#2212)
cmikeh2 Sep 23, 2022
76de924
fix zero docs (#2350)
jeffra Sep 23, 2022
9932643
Inference profiling updates/fixes (#2348) (#2349)
awan-10 Sep 23, 2022
9aa7b63
Kernel Data Conversion Utility (#2327)
cmikeh2 Sep 26, 2022
f210256
Add Onebit Optimzers in __init__ (#2340)
l4d2boomer Sep 26, 2022
2b1b0d2
docs(mixture-of-experts-inference): fix typo in tuto (#2345)
jqueguiner Sep 26, 2022
6ef16de
download cifar to blob storage (#2342)
mrwyattii Sep 26, 2022
9df604b
Refactor gptj_residual_add kernels for better readability (#2358)
arashb Sep 26, 2022
70e883a
Updated issue templates (#2363)
jeffra Sep 26, 2022
8e8c866
Update issue templates
jeffra Sep 26, 2022
3486afb
fix cuda invalid config error in dequant kernel (#2362)
GuanhuaWang Sep 27, 2022
b450da4
Add missing pytest fixture scope (#2353)
arashb Sep 27, 2022
79692af
Extend residual_add kernel tests to conver pre_attn_norm (#2354)
arashb Sep 27, 2022
e14d40e
Refactor fused_bias_residual kernels for better readability (#2356)
arashb Sep 27, 2022
eed4032
Capture error message during sweep tests (#2351)
molly-smith Sep 27, 2022
b609a29
fix an exception when recursively casting dicts to fp16 (#2370)
mjksmith Sep 29, 2022
ff42743
Refactor remaining distributed tests (#2216)
mrwyattii Oct 4, 2022
0a2ae2e
Fix the MLP output tensor's shape (#2380)
arashb Oct 5, 2022
f5a8348
allow building with latest CUDA (11.8), it is backwards compatible (#…
Thomas-MMJ Oct 5, 2022
6f3dec6
pin transformers version for unit tests (#2402)
mrwyattii Oct 7, 2022
46a886c
Change type to tuple in replace_wo_policy isinstance check (#2387)
lekurile Oct 7, 2022
f4a92a1
Checkpoint backwards-compatbility workaround (#2384)
tjruwase Oct 9, 2022
c300132
Add predicated global load (#2373)
cmikeh2 Oct 11, 2022
b8b98c6
MII blog post (#2418)
jeffra Oct 12, 2022
d0c6b29
Fix figure reference (#2419)
awan-10 Oct 12, 2022
ab77677
[docs] update news items
jeffra Oct 12, 2022
5a16213
[docs] add mii repo link
jeffra Oct 12, 2022
3db0b5e
Add SLURM Multinode Runner (#2404)
dashstander Oct 13, 2022
d5d10b0
Fix issue with corrupted output on long generation for GPT (#2359)
andrewchernyh Oct 13, 2022
3789dc5
MII blog title update on Readme
samyam Oct 13, 2022
843fd4c
DeepSpeed-MII title change in website
samyam Oct 13, 2022
cd3a709
Fix GPT Neo-X multi-gpu inference (#2401)
andrewchernyh Oct 13, 2022
1ffcfeb
MII-Public and MII-Azure subheading in mii post
samyam Oct 13, 2022
1c39fe1
CI fixes related to triton (#2422)
jeffra Oct 13, 2022
d716537
[docs] update mii blog title (#2423)
jeffra Oct 13, 2022
ec13da6
add SD injection policy (#2381)
jeffra Oct 13, 2022
537e858
fix checkpoint loading when it is a dictionary (#2425)
RezaYazdaniAminabadi Oct 14, 2022
5eafb8c
Make error regex more generic in collect_results.py (#2415)
molly-smith Oct 14, 2022
cfead55
fixes #2389 (#2411)
Oct 14, 2022
34fb6d1
Fix for inference gpt-j test (#2430)
mrwyattii Oct 17, 2022
906b4a0
Fixing bug 2361 (#2410)
jomayeri Oct 18, 2022
799120e
Universal checkpoint for zero stage 1 (#2284)
tjruwase Oct 18, 2022
1b7c679
only add deps if extra is explictly called (#2432)
jeffra Oct 18, 2022
b2a724e
Add TestInjectionPolicy inference unittest class for testing custom i…
lekurile Oct 18, 2022
99fde3b
[memory estimators] new config args sync (#2431)
stas00 Oct 20, 2022
b8fb9c3
parallelize writing of layer checkpoint files across data parallel in…
adammoody Oct 21, 2022
877a881
Fix broken link to DeepSpeed Megatron fork (#2440)
lekurile Oct 21, 2022
a524864
bump to 0.7.5
jeffra Oct 21, 2022
7d11363
Fix Bug #2319 (#2438)
jomayeri Oct 24, 2022
5d1f595
update pytorch pool operator function signiture (#2443)
cli99 Oct 25, 2022
b85eb3b
Fix build issues on Windows (#2428)
eltonzheng Oct 26, 2022
8da0238
rollback ds config changes (#2395)
cli99 Oct 26, 2022
e772f16
Use CUDA events for inference model profiling (#2371)
mrwyattii Oct 26, 2022
3b3ba3c
Fixing a mismatch in basic adam test. (#2447)
jomayeri Oct 26, 2022
be4ffb8
Reduction Kernel Utility (#2436)
cmikeh2 Oct 27, 2022
3432c74
deepspeed/launcher/launch.py: add option '--enable_each_rank_log logd…
guoyejun Oct 28, 2022
825f9d4
Fixes for various CI problems (#2457)
mrwyattii Nov 2, 2022
10e9d04
Cache Allocation and Softmax Fixes (#2433)
cmikeh2 Nov 2, 2022
39bdc14
fixing the checkpoint loading at inference-engine (#2429)
RezaYazdaniAminabadi Nov 2, 2022
35458da
Create a new folder structure to isolate model-specific code in DS (#…
awan-10 Nov 4, 2022
f74ee31
don't gather partitioned activations for mp size 1 (#2454)
guoyejun Nov 4, 2022
4a06ecf
Updating autotune json default in docs. (#2476)
jomayeri Nov 4, 2022
ffb6d98
Added MLFLOW environment variables for logging metrics within trainig…
savitamittal1 Nov 5, 2022
a47c3e0
fix accelerate link (#2481)
kyoto7250 Nov 7, 2022
9cfcf74
Add correct memory-allocation at DeepSpeed-Attention (#2474)
RezaYazdaniAminabadi Nov 8, 2022
521d329
Fix CI issues related to cupy install (#2483)
mrwyattii Nov 8, 2022
a63cb07
[docs] add SD tutorial to news
jeffra Nov 9, 2022
d2d1b4c
[docs] add SD tutorial to deepspeed.ai news
jeffra Nov 9, 2022
6f77da1
Add `scale_attn_by_inverse_layer_idx` feature (#2486)
hyunwoongko Nov 9, 2022
e7e7595
Stable Diffusion Enhancements (#2491)
cmikeh2 Nov 10, 2022
3ca9878
stage_1_and_2.py: no allreduce needed when mp size is 1 (#2494)
guoyejun Nov 10, 2022
ee39187
Make bf16_optimizer work for non pipeline (#2470)
tjruwase Nov 10, 2022
be5ec50
Fix nightly CI tests (#2493)
mrwyattii Nov 10, 2022
f2710bb
Make data contiguous before the inplace reshape-copy_ function (#2489)
lokoppakmsft Nov 11, 2022
06e00f6
Fix typos: deepseed -> deepspeed (#2499)
jinyouzhi Nov 13, 2022
a4ceabb
bump to 0.7.6
jeffra Nov 14, 2022
b5d18a6
DeepSpeed inference config. (#2459) (#2472)
awan-10 Nov 15, 2022
43bf035
Update docs to autogenerate pydantic config model docs (#2509)
mrwyattii Nov 15, 2022
d40a15f
Add max_tokens alias to max_out_tokens arg to maintain backwards comp…
lekurile Nov 15, 2022
78d4ca1
Deepspeed quantization library v0.1 (#2450)
lokoppakmsft Nov 17, 2022
e59f805
Fix backward compatibility for InferenceConfig (#2516)
mrwyattii Nov 17, 2022
fe67854
Add missing Inference sub-configs (#2518)
mrwyattii Nov 17, 2022
7ce371b
Add note about nvcc/hipcc requirement (#2519)
jeffra Nov 17, 2022
0b26532
Update codeowners (#2525)
jeffra Nov 18, 2022
30c8d8a
Initial dequant library implementation (#2521)
cmikeh2 Nov 19, 2022
2110552
Fixes for torch 1.14 due to new torch.numel return type (#2522)
jeffra Nov 21, 2022
57e0a55
Ensure is initialized for SD (#2534)
cmikeh2 Nov 22, 2022
8b4318b
Make DS-Inference config readable from JSON (#2537)
mrwyattii Nov 22, 2022
c5ee27f
Add MII tests (#2533)
mrwyattii Nov 22, 2022
4abf637
Remove mutable default parameter in init_inference() (#2540)
aphedges Nov 23, 2022
5df1eea
Change Where DS/Triton is Used in Stable Diffusion (#2536)
cmikeh2 Nov 23, 2022
90ae688
Pass down the new DS inference config to replace_transformer_layer. (…
awan-10 Nov 23, 2022
21c2802
Adding Gradient Accumulation Data Type Config (#2512)
jomayeri Nov 28, 2022
340fc0c
Report progress at gradient accumulation boundary (#2553)
ShijieZZZZ Nov 29, 2022
abe4fc6
encoded ds config into command line argument when launching child pro…
cli99 Nov 29, 2022
c5f8585
add missing moe deprecated fields to inference config (#2556)
mrwyattii Nov 29, 2022
ffcf384
Abstract accelerator (step 1) (#2504)
delock Nov 30, 2022
aeda7f9
Fix invalid check of recorded parameter orders in zero stage3. (#2550)
inkcherry Nov 30, 2022
c77d42d
bump to 0.7.7
jeffra Dec 1, 2022
bbe030c
docs: Update the recent url for Megatron-LM (#2564)
rapsealk Dec 2, 2022
d4cab2c
use get_global_rank if available (#2567)
jeffra Dec 2, 2022
2d8f3f5
Add Determined to open-source DL frameworks (#2573)
sirredbeard Dec 5, 2022
0693883
Support fp32 gradaccum for bf16 model (#2566)
delock Dec 6, 2022
b841628
Drop Maxwell Support (#2574)
cmikeh2 Dec 6, 2022
35b350b
Fix quantized-inference & Add generic support of checkpoint loading (…
RezaYazdaniAminabadi Dec 6, 2022
731965d
Fix MegatronLayerPolicy to have megatron_v2=True (#2579)
lekurile Dec 7, 2022
18d55e5
Update barrier and reduce_scatter_base to conform to PyTorch signatur…
Quentin-Anthony Dec 7, 2022
591744e
Support N-dimension input in quantization kernel (#2575)
lokoppakmsft Dec 8, 2022
ccb8eb8
Add checkpoint sharding unit tests (#2561)
mrwyattii Dec 8, 2022
377c770
Updating docs README (#2587)
jomayeri Dec 9, 2022
18713c6
Updating API docs (#2586)
jomayeri Dec 9, 2022
35eabb0
Fix issues w. python 3.6 + add py-version checks to CI (#2589)
jeffra Dec 9, 2022
2076bf2
get mask token from tokenizer (#2592)
jeffra Dec 10, 2022
2600db5
bump to 0.7.8
jeffra Dec 12, 2022
ef86937
DeepSpeed Data Efficiency Library (#2585)
conglongli Dec 13, 2022
acde873
fix blog link (#2600)
conglongli Dec 13, 2022
7425a36
Migrate ops tests to new inference_ops marker (#2599)
cmikeh2 Dec 13, 2022
3a3dfe6
Move layer norm to new schedule (#2590)
lokoppakmsft Dec 13, 2022
384f17b
[deepspeed/autotuner] Bug fix for binary search for batch size (#2162)
rahilbathwal5 Dec 14, 2022
2e596e6
add fix for older pydantic versions (#2611)
mrwyattii Dec 15, 2022
7e2103f
Use rocm/pytorch:latest (#2613)
jithunnair-amd Dec 15, 2022
b08cf41
skip torch.zeros and tensor.copy_ when model parallel is not used (#2…
guoyejun Dec 16, 2022
d0dbc95
call empty_cache to really free up GPU memory as described in comment…
guoyejun Dec 16, 2022
503706a
Remove GatheredParameters context from replace_with_policy (#2591)
lekurile Dec 16, 2022
0f0e38c
fixes #2498 (#2603)
Dec 17, 2022
a25c31b
Update AVX512 Detection (#2621)
cmikeh2 Dec 17, 2022
df985fa
Add Megatron CI workflow (#2614)
mrwyattii Dec 18, 2022
5676f5e
[inference] check for unsupported model generate args (#2627)
jeffra Dec 19, 2022
8c56c25
[launcher] parse hostfile via regex and added error checks (#2626)
jeffra Dec 19, 2022
6fff50f
Unit tests setup own venv (#2628)
mrwyattii Dec 20, 2022
11f5dab
add enable_each_rank_log to deepspeed/launcher/runner.py (#2571)
inkcherry Dec 21, 2022
6273dff
Fix typo in autotuner.py (#2639)
eltociear Dec 21, 2022
a298a43
[zero-3] Handle forward parameter return correctly in nested cases (#…
samyam Dec 23, 2022
bb68c52
[inference] ds-attention refactor w.r.t. ops (#2623)
jeffra Dec 23, 2022
e0aa84c
Fix issue w. bloom when changing tp size (#2645)
jeffra Dec 23, 2022
6375cb3
fix assertion error in zero stage 3 (#2647)
GuanhuaWang Dec 23, 2022
d9b788d
tweaks to ds-attn, distilbert policy, and mup (#2649)
jeffra Dec 28, 2022
f2ea7a3
[doc] fix `min_loss_scale` default (#2660)
stas00 Jan 3, 2023
a091bc2
[launcher] fail gracefully if hostname -i doesn't work as expected (#…
jeffra Jan 3, 2023
95d9a1b
Fix Opt injection (#2541)
RezaYazdaniAminabadi Jan 6, 2023
9548d48
Abstract accelerator (step 2) (#2560)
delock Jan 7, 2023
97deaae
Remove unnecessary device synchronization for stage 2 (#2500)
li-yi-dong Jan 8, 2023
323c266
[Bug Fixed] use torch.cuda.is_available() (#2661)
wkcn Jan 9, 2023
f30a030
[fp16] lower initial_scale_power (#2663)
stas00 Jan 9, 2023
be6d19f
fix Tensor contiguous bug in model_compression (#2671)
xiaoxiawu-microsoft Jan 9, 2023
c702b64
[inference] ds-mlp refactor w.r.t. ops (#2668)
jeffra Jan 10, 2023
62c071e
real_accelerator validation check for both accelerator and deepspeed.…
delock Jan 10, 2023
89da037
remove duplicated code in ZeRO (#2655)
wkcn Jan 10, 2023
a3d7f10
Add mlflow logging for aml (#2495)
cassieesvelt Jan 11, 2023
cf9e433
fix import path to op_builder (#2687)
tohtana Jan 11, 2023
e7c1402
Pass training flag to forward call from Eval (#2604)
lokoppakmsft Jan 11, 2023
aef8a85
Extend quantization utils features (#2683)
lokoppakmsft Jan 12, 2023
217cc07
[GatheredParameters] add support for any iterator (#2664)
stas00 Jan 13, 2023
c9c6ab9
fix for latest diffusers (#2699)
mrwyattii Jan 13, 2023
cd271a4
exclude benchmarks during install (#2698)
jeffra Jan 13, 2023
fe728e3
using correct loss scale in zero step (#2695)
jomayeri Jan 14, 2023
e4ba722
non-MoE stage 1 requires CG disabled (#2703)
jeffra Jan 17, 2023
35575bc
remove print side effect from importing deepspeed (#2704)
jeffra Jan 17, 2023
bf6b980
ZeRO3 handling frozen weights] (#2653)
tjruwase Jan 17, 2023
7d0e427
bump to 0.8.1
jeffra Jan 17, 2023
3f210c9
CUDA optional deepspeed ops (#2507)
tjruwase Jan 17, 2023
df2495c
remove master branch from CI triggers (#2712)
mrwyattii Jan 18, 2023
0b549ad
[install] only add deepspeed pkg at install (#2714)
jeffra Jan 18, 2023
23e5133
update for lm-eval==0.3.0 (#2713)
mrwyattii Jan 18, 2023
8d87c89
BF16 optimizer for BF16+ZeRO Stage 1 (#2706)
jomayeri Jan 18, 2023
8df50a2
fix typo (#2718)
mrwyattii Jan 18, 2023
867da30
Inference Refactor (replace_with_policy, model_implementations) (#2554)
awan-10 Jan 19, 2023
34a1168
Change zero_grad() argument to match pytorch (#2741)
loadams Jan 24, 2023
d59b572
Automatic tensor parallelism v2 (#2670)
molly-smith Jan 24, 2023
4be8df7
fixing optimizer sanity check (#2742)
jomayeri Jan 25, 2023
ddd48b3
[GatheredParameters] fix memory leak (#2665)
stas00 Jan 26, 2023
98cc35b
Abstract accelerator (step 3) (#2677)
delock Jan 26, 2023
d4bfae4
Fix autotuning so that it records Floating Point Operations per secon…
dashstander Jan 26, 2023
30d3f5d
fix a mispelled attribute (#2750)
stas00 Jan 26, 2023
a60e31a
[zero] remove misleading dtype log (#2732)
jeffra Jan 26, 2023
0b06e0c
Fix softmax backward (#2709)
RezaYazdaniAminabadi Jan 26, 2023
cc3d7cb
Skip test_bias_gelu unit test if torch < 1.12 (#2754)
lekurile Jan 27, 2023
258d283
Add environment variable to make nvcc compilation more verbose (#2759)
cmikeh2 Jan 28, 2023
8d3b42c
Bing/formatting correction (#2764)
xiexbing Jan 30, 2023
1db4ade
Add links to new azureML examples (#2756)
cassieesvelt Jan 31, 2023
8647753
Fix hardcoded instances to fp16 in optimizer creation log messages to…
loadams Jan 31, 2023
d923f7c
Refactor/Pydantify monitoring config (#2640)
mrwyattii Jan 31, 2023
02e95e6
Pin minimum `packaging` requirement (#2771)
carmocca Jan 31, 2023
ef6a958
Fix for diffusers v0.12.0 (#2753)
mrwyattii Jan 31, 2023
e2a31d8
some fix in flops_profiler (#2068)
lucasleesw Feb 1, 2023
b5750b6
fix upsample flops compute by skipping unused kargs (#2773)
cli99 Feb 1, 2023
c5b983e
Fix broken kernel inject bug (#2776)
molly-smith Feb 2, 2023
2c6e819
Fix Checkpoint-loading with Meta-tensor (#2781)
RezaYazdaniAminabadi Feb 3, 2023
4079077
add support for hjson config files (#2783)
mrwyattii Feb 3, 2023
9f41ffe
Reset KV-cache at the beginning of text-generation (#2669)
RezaYazdaniAminabadi Feb 3, 2023
0a73e6e
Container param cleanup + remove qkv_merging (#2780)
lekurile Feb 3, 2023
7d9fae4
Common location to install libaio-dev (#2779)
tjruwase Feb 4, 2023
f376dae
Fixing broken link to azureml-examples recipes (#2795)
rtanase Feb 6, 2023
d323abd
remove outdated comment (#2786)
stas00 Feb 6, 2023
c9b0888
Enable page-locked tensors without CUDA (#2775)
tjruwase Feb 7, 2023
10f3c30
Add container load checkpoint error reporting + refactor (#2792)
lekurile Feb 7, 2023
4af1f76
Add user defined launcher args for PDSH launcher (#2804)
loadams Feb 8, 2023
d038dbd
Fix Slurm launcher user args (#2806)
loadams Feb 9, 2023
349f845
Handle hanged tests in CI (#2808)
mrwyattii Feb 11, 2023
93e5289
Fix inference CI device error (#2824)
mrwyattii Feb 14, 2023
d3f3019
Fix permissions issue with pip upgrade (#2823)
mrwyattii Feb 14, 2023
639aa7b
Fix cpu-only CI hangs (#2825)
mrwyattii Feb 14, 2023
cc1054d
Fix Pipeline Parallel resize unit test (#2833)
mrwyattii Feb 15, 2023
46784cb
Fix auto TP for duplicate modules with different gems (#2784)
molly-smith Feb 15, 2023
e4b3b61
Refactor DS inference API. No longer need replace_method. (#2831)
awan-10 Feb 15, 2023
fd1449c
Port Reza's INT8-quantization fix to container architecture (#2725)
lekurile Feb 16, 2023
5b7413a
Fix gpt-Neox rotary embedding implementation (#2782)
RezaYazdaniAminabadi Feb 16, 2023
6c85fe6
fix for cpu-only tests (#2849)
mrwyattii Feb 17, 2023
aee250b
bump to 0.8.2
jeffra Feb 17, 2023
3ea272e
add auto-generated PR workflow (#2822)
mrwyattii Feb 17, 2023
8be8c01
fix typo in autosync workflow (#2850)
mrwyattii Feb 17, 2023
b4d40e3
Fix example command when building wheel with dev version specified (…
loadams Feb 21, 2023
3253e7d
Create tensor parallelism blog/tutorial (#2766)
molly-smith Feb 21, 2023
7c99def
Data efficiency library update (#2866)
conglongli Feb 21, 2023
81b4d5d
Make z3 respect comm dtype (#2807)
tjruwase Feb 22, 2023
859d7c9
Automatic Tensor Parallelism Blog Links (#2877)
molly-smith Feb 23, 2023
7e77cf7
Check device count before running dist tests (#2799)
HeyangQin Feb 23, 2023
b47c592
AutoTP tutorial web formatting and news (#2883)
molly-smith Feb 23, 2023
d3de737
Remove deprecated `torch._six` imports (#2863)
yasyf Feb 24, 2023
8710f05
Reduce I/O size (#2814)
tjruwase Feb 24, 2023
da84e60
add missing license info to top of all source code (#2889)
jeffra Feb 27, 2023
541e423
Enable tensor fragments for zero 2 & 3 (#2727)
tjruwase Feb 28, 2023
f1d2a15
better eval sampler (#2907)
mayank31398 Feb 28, 2023
dc01cee
using container when loading inference checkpoints (#2875)
HeyangQin Feb 28, 2023
9886d6d
Fix CPUAdam for when `vendor_id_raw` is not provided (#2836)
FarzanT Feb 28, 2023
17fa087
Always convert input mask to half (#2851)
molly-smith Feb 28, 2023
91d7090
Fixes `AttributeError` in #2853 (#2854)
saforem2 Mar 1, 2023
8d53ac0
Add MPICH Multinode Runner (#2839)
inkcherry Mar 1, 2023
4ae3a3d
TP unsupported models and assertions (#2810)
molly-smith Mar 1, 2023
2ede0d9
AutoTP Assert Kernel Injection Support (#2939)
molly-smith Mar 6, 2023
87eaf8f
Check for local CUDA graphs when enable_cuda_graph=True (#2941)
lekurile Mar 7, 2023
80d8fcb
Improve overflow handling (#2944)
tjruwase Mar 7, 2023
0acf7e9
[RFC] add device abstraction to allow other device than CUDA be used …
delock Mar 7, 2023
db15ef5
deepspeed.init_distributed() support for TCP protocols (#2905)
noabauma Mar 7, 2023
d58b4df
bump to 0.8.3
jeffra Mar 7, 2023
6379def
bug fix for skipping mbs (#2171)
rahilbathwal5 Mar 8, 2023
58a4a4d
Fix issue between our abstract accelerator and colossalai's version o…
jeffra Mar 8, 2023
457850d
[zero] prevent poor configs from running w. zero-offload (#2971)
jeffra Mar 9, 2023
1622ff0
Merge branch 'master' into update_deeperspeed_final
Quentin-Anthony Mar 9, 2023
7f2d926
Merge branch 'main' into update_deeperspeed_final
Quentin-Anthony Mar 9, 2023
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
310 changes: 155 additions & 155 deletions .clang-format
Original file line number Diff line number Diff line change
@@ -1,155 +1,155 @@
---
# Refer to the following link for the explanation of each params:
# http://releases.llvm.org/8.0.0/tools/clang/docs/ClangFormatStyleOptions.html
Language: Cpp
# BasedOnStyle: Google
AccessModifierOffset: -4
AlignAfterOpenBracket: Align
AlignConsecutiveAssignments: false
AlignConsecutiveDeclarations: false
AlignEscapedNewlines: Left
AlignOperands: true
AlignTrailingComments: true
AllowAllParametersOfDeclarationOnNextLine: false
AllowShortBlocksOnASingleLine: true
AllowShortCaseLabelsOnASingleLine: true
AllowShortFunctionsOnASingleLine: All
AllowShortIfStatementsOnASingleLine: true
AllowShortLoopsOnASingleLine: true
# This is deprecated
AlwaysBreakAfterDefinitionReturnType: None
AlwaysBreakAfterReturnType: None
AlwaysBreakBeforeMultilineStrings: true
AlwaysBreakTemplateDeclarations: true
BinPackArguments: false
BinPackParameters: false
BraceWrapping:
AfterClass: false
AfterControlStatement: false
AfterEnum: false
AfterFunction: false
AfterNamespace: false
AfterObjCDeclaration: false
AfterStruct: false
AfterUnion: false
AfterExternBlock: false
BeforeCatch: false
BeforeElse: false
IndentBraces: false
# disabling the below splits, else, they'll just add to the vertical length of source files!
SplitEmptyFunction: false
SplitEmptyRecord: false
SplitEmptyNamespace: false
BreakBeforeBinaryOperators: None
BreakBeforeBraces: WebKit
BreakBeforeInheritanceComma: false
BreakInheritanceList: BeforeColon
BreakBeforeTernaryOperators: true
BreakConstructorInitializersBeforeComma: false
BreakConstructorInitializers: BeforeColon
BreakAfterJavaFieldAnnotations: false
BreakStringLiterals: true
ColumnLimit: 100
CommentPragmas: '^ IWYU pragma:'
CompactNamespaces: false
ConstructorInitializerAllOnOneLineOrOnePerLine: true
# Kept the below 2 to be the same as `IndentWidth` to keep everything uniform
ConstructorInitializerIndentWidth: 4
ContinuationIndentWidth: 4
Cpp11BracedListStyle: true
DerivePointerAlignment: false
DisableFormat: false
ExperimentalAutoDetectBinPacking: false
FixNamespaceComments: true
ForEachMacros:
- foreach
- Q_FOREACH
- BOOST_FOREACH
IncludeBlocks: Preserve
IncludeCategories:
- Regex: '^<ext/.*\.h>'
Priority: 2
- Regex: '^<.*\.h>'
Priority: 1
- Regex: '^<.*'
Priority: 2
- Regex: '.*'
Priority: 3
IncludeIsMainRegex: '([-_](test|unittest))?$'
IndentCaseLabels: true
IndentPPDirectives: None
IndentWidth: 4
IndentWrappedFunctionNames: false
JavaScriptQuotes: Leave
JavaScriptWrapImports: true
KeepEmptyLinesAtTheStartOfBlocks: false
MacroBlockBegin: ''
MacroBlockEnd: ''
MaxEmptyLinesToKeep: 1
NamespaceIndentation: None
ObjCBinPackProtocolList: Never
ObjCBlockIndentWidth: 4
ObjCSpaceAfterProperty: false
ObjCSpaceBeforeProtocolList: true
PenaltyBreakAssignment: 4
PenaltyBreakBeforeFirstCallParameter: 1
PenaltyBreakComment: 300
PenaltyBreakFirstLessLess: 120
PenaltyBreakString: 1000
PenaltyBreakTemplateDeclaration: 10
PenaltyExcessCharacter: 1000000
PenaltyReturnTypeOnItsOwnLine: 200
PointerAlignment: Left
RawStringFormats:
- Language: Cpp
Delimiters:
- cc
- CC
- cpp
- Cpp
- CPP
- 'c++'
- 'C++'
CanonicalDelimiter: ''
- Language: TextProto
Delimiters:
- pb
- PB
- proto
- PROTO
EnclosingFunctions:
- EqualsProto
- EquivToProto
- PARSE_PARTIAL_TEXT_PROTO
- PARSE_TEST_PROTO
- PARSE_TEXT_PROTO
- ParseTextOrDie
- ParseTextProtoOrDie
CanonicalDelimiter: ''
BasedOnStyle: google
# Enabling comment reflow causes doxygen comments to be messed up in their formats!
ReflowComments: true
SortIncludes: true
SortUsingDeclarations: true
SpaceAfterCStyleCast: false
SpaceAfterTemplateKeyword: true
SpaceBeforeAssignmentOperators: true
SpaceBeforeCpp11BracedList: false
SpaceBeforeCtorInitializerColon: true
SpaceBeforeInheritanceColon: true
SpaceBeforeParens: ControlStatements
SpaceBeforeRangeBasedForLoopColon: true
SpaceInEmptyParentheses: false
SpacesBeforeTrailingComments: 2
SpacesInAngles: false
SpacesInContainerLiterals: true
SpacesInCStyleCastParentheses: false
SpacesInParentheses: false
SpacesInSquareBrackets: false
Standard: Cpp11
StatementMacros:
- Q_UNUSED
- QT_REQUIRE_VERSION
# Be consistent with indent-width, even for people who use tab for indentation!
TabWidth: 4
UseTab: Never
---
# Refer to the following link for the explanation of each params:
# http://releases.llvm.org/8.0.0/tools/clang/docs/ClangFormatStyleOptions.html
Language: Cpp
# BasedOnStyle: Google
AccessModifierOffset: -4
AlignAfterOpenBracket: Align
AlignConsecutiveAssignments: false
AlignConsecutiveDeclarations: false
AlignEscapedNewlines: Left
AlignOperands: true
AlignTrailingComments: true
AllowAllParametersOfDeclarationOnNextLine: false
AllowShortBlocksOnASingleLine: true
AllowShortCaseLabelsOnASingleLine: true
AllowShortFunctionsOnASingleLine: All
AllowShortIfStatementsOnASingleLine: true
AllowShortLoopsOnASingleLine: true
# This is deprecated
AlwaysBreakAfterDefinitionReturnType: None
AlwaysBreakAfterReturnType: None
AlwaysBreakBeforeMultilineStrings: true
AlwaysBreakTemplateDeclarations: true
BinPackArguments: false
BinPackParameters: false
BraceWrapping:
AfterClass: false
AfterControlStatement: false
AfterEnum: false
AfterFunction: false
AfterNamespace: false
AfterObjCDeclaration: false
AfterStruct: false
AfterUnion: false
AfterExternBlock: false
BeforeCatch: false
BeforeElse: false
IndentBraces: false
# disabling the below splits, else, they'll just add to the vertical length of source files!
SplitEmptyFunction: false
SplitEmptyRecord: false
SplitEmptyNamespace: false
BreakBeforeBinaryOperators: None
BreakBeforeBraces: WebKit
BreakBeforeInheritanceComma: false
BreakInheritanceList: BeforeColon
BreakBeforeTernaryOperators: true
BreakConstructorInitializersBeforeComma: false
BreakConstructorInitializers: BeforeColon
BreakAfterJavaFieldAnnotations: false
BreakStringLiterals: true
ColumnLimit: 100
CommentPragmas: '^ IWYU pragma:'
CompactNamespaces: false
ConstructorInitializerAllOnOneLineOrOnePerLine: true
# Kept the below 2 to be the same as `IndentWidth` to keep everything uniform
ConstructorInitializerIndentWidth: 4
ContinuationIndentWidth: 4
Cpp11BracedListStyle: true
DerivePointerAlignment: false
DisableFormat: false
ExperimentalAutoDetectBinPacking: false
FixNamespaceComments: true
ForEachMacros:
- foreach
- Q_FOREACH
- BOOST_FOREACH
IncludeBlocks: Preserve
IncludeCategories:
- Regex: '^<ext/.*\.h>'
Priority: 2
- Regex: '^<.*\.h>'
Priority: 1
- Regex: '^<.*'
Priority: 2
- Regex: '.*'
Priority: 3
IncludeIsMainRegex: '([-_](test|unittest))?$'
IndentCaseLabels: true
IndentPPDirectives: None
IndentWidth: 4
IndentWrappedFunctionNames: false
JavaScriptQuotes: Leave
JavaScriptWrapImports: true
KeepEmptyLinesAtTheStartOfBlocks: false
MacroBlockBegin: ''
MacroBlockEnd: ''
MaxEmptyLinesToKeep: 1
NamespaceIndentation: None
ObjCBinPackProtocolList: Never
ObjCBlockIndentWidth: 4
ObjCSpaceAfterProperty: false
ObjCSpaceBeforeProtocolList: true
PenaltyBreakAssignment: 4
PenaltyBreakBeforeFirstCallParameter: 1
PenaltyBreakComment: 300
PenaltyBreakFirstLessLess: 120
PenaltyBreakString: 1000
PenaltyBreakTemplateDeclaration: 10
PenaltyExcessCharacter: 1000000
PenaltyReturnTypeOnItsOwnLine: 200
PointerAlignment: Left
RawStringFormats:
- Language: Cpp
Delimiters:
- cc
- CC
- cpp
- Cpp
- CPP
- 'c++'
- 'C++'
CanonicalDelimiter: ''
- Language: TextProto
Delimiters:
- pb
- PB
- proto
- PROTO
EnclosingFunctions:
- EqualsProto
- EquivToProto
- PARSE_PARTIAL_TEXT_PROTO
- PARSE_TEST_PROTO
- PARSE_TEXT_PROTO
- ParseTextOrDie
- ParseTextProtoOrDie
CanonicalDelimiter: ''
BasedOnStyle: google
# Enabling comment reflow causes doxygen comments to be messed up in their formats!
ReflowComments: true
SortIncludes: true
SortUsingDeclarations: true
SpaceAfterCStyleCast: false
SpaceAfterTemplateKeyword: true
SpaceBeforeAssignmentOperators: true
SpaceBeforeCpp11BracedList: false
SpaceBeforeCtorInitializerColon: true
SpaceBeforeInheritanceColon: true
SpaceBeforeParens: ControlStatements
SpaceBeforeRangeBasedForLoopColon: true
SpaceInEmptyParentheses: false
SpacesBeforeTrailingComments: 2
SpacesInAngles: false
SpacesInContainerLiterals: true
SpacesInCStyleCastParentheses: false
SpacesInParentheses: false
SpacesInSquareBrackets: false
Standard: Cpp11
StatementMacros:
- Q_UNUSED
- QT_REQUIRE_VERSION
# Be consistent with indent-width, even for people who use tab for indentation!
TabWidth: 4
UseTab: Never
43 changes: 43 additions & 0 deletions .github/ISSUE_TEMPLATE/compression_bug_report.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,43 @@
---
name: Bug report (compression)
about: Create a DeepSpeed compression related issue to help us improve
title: "[BUG]"
labels: bug,compression
assignees: ''

---

**Describe the bug**
A clear and concise description of what the bug is.

**To Reproduce**
Steps to reproduce the behavior:
1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

**Expected behavior**
A clear and concise description of what you expected to happen.

**ds_report output**
Please run `ds_report` to give us details about your setup.

**Screenshots**
If applicable, add screenshots to help explain your problem.

**System info (please complete the following information):**
- OS: [e.g. Ubuntu 18.04]
- GPU count and types [e.g. two machines with x8 A100s each]
- Interconnects (if applicable) [e.g., two machines connected with 100 Gbps IB]
- Python version
- Any other relevant info about your setup

**Launcher context**
Are you launching your experiment with the `deepspeed` launcher, MPI, or something else?

**Docker context**
Are you using a specific docker image that you can share?

**Additional context**
Add any other context about the problem here.
20 changes: 20 additions & 0 deletions .github/ISSUE_TEMPLATE/feature_request.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,20 @@
---
name: Feature request
about: Suggest an idea for this project
title: "[REQUEST]"
labels: enhancement
assignees: ''

---

**Is your feature request related to a problem? Please describe.**
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

**Describe the solution you'd like**
A clear and concise description of what you want to happen.

**Describe alternatives you've considered**
A clear and concise description of any alternative solutions or features you've considered.

**Additional context**
Add any other context or screenshots about the feature request here.
Loading