1.2.x cherries 🍒 #6083
Description
@PyTorchLightning/core-contributors Please update this issue with any commits/PRs that should make it into 1.2.x.
Each cherry is to be git cherry-pick-ed into the appropriate release
1.2.2 (#6304)
40d5a9d (#5931)
3df02b8 (#6072): feature, required for #6073
1b498d1 (#6075)
925f082 (#6147)
dd2f5a0 (#6197)
651c25f (#6216)
efda48f (#6262)
412a7d8 (#6267)
8001987 (#6297)
1.2.3 (#6431)
4a8422c (#6073)
4f90455 (#6076)
826375e (#6136): blocked by (#4945)
facfda8 (#6139): blocked by (#4945)
5d7388d (#6177)
a6c98c4 (#6221)
2ec67a4 (#6260)
1aac481 (#6272)
484dce1 (#6296)
8001987 (#6302)
7acbd65 (#6305)
e038e74 (#6323)
46540ee (#6326)
39231ae (#6330)
e848542 (#6357)
85c8074 (#6363)
2a3ab67 (#6365)
718074b (#6372)
0ec7a23 (#6399)
1.2.4 (#6545)
248a8e8 (#6324)
523c59b (#4688)
b2bcad1 (#5968)
c53edce (#6098)
fc6d402 (#6380)
0544efd (#6410)
9eded7f (#6419)
c81b2a8 (#6438)
1c013b4 (#6460)
079fe9b (#6476)
02fa32b (#6511)
156847b (#6516)
ea36ee3 (#6513)
6a14146 (#6537)
0f07eaf (#6510)
last sha scrutinized: 0f07eaf
1.2.5 (#6646)
bc577ca (#6275)
8cd75a4 (#6434)
6453091 (#6530)
555a6fe (#6540)
634d831 (#6565)
87c03b1 (#6576)
8853a36 (#6579)
983a888 (#6587)
1fae10a (#6590)
cb59039 (#6600)
last sha scrutinized: 1fae10a
1.2.6 (#6729)
615b2f7 (#6398)
9be092d (#6498)
64d0fa4 (#6524)
3b72bcc (#6536)
b8ef52b (#6625)
e62c7c7 (#6632)
2cbdc01 (#6654)
741c452 (#6657)
bc61361 (#6682)
dcf6e4e (#6686)
3a4c424 (#6719)
last sha scrutinized: f79a13e
1.2.7 (#6850)
1bd5f36 (#5952)
6dc1078 (#6588)
3c86193 (#6728)
89b5326 (#6730)
bb92754 (#6741)
9876df1 (#6743)
13f67ad (#6767)
3e3175d (#6781)
f581411 (#6784)
22a266d (#6794)
f8a3798 (#6801)
cf8e828 (#6816)
7f91c5e (#6822)
264aa68 (#6828)
c3da7f5 (#6836)
eafec7d (#6838)
last sha scrutinized: f581411
1.2.8 (#6983)
23e8dff [CI] Drop brew update (#6985)
b37b58a Fix Checkpoint issue when using Horovod distributed backend (#6958)
da1ac3a update docker base on PT 1.7 (#6931)
1b3e4f9 Fix sync_dist for tpus (#6950)
80c5293 fix self.device access in DataParallel (#6414)
fe0d088 Fix ShardedDataParallel has no attribute require_backward_grad_sync (#6915)
5552503 Fix TPU Spawn gather (#6896)
9c9e2a0 fix gpus default for Trainer.add_argparse_args (#6898)
90e37ba fix exception raising (#6901)
742c48e [Fix] Ensure we set the eval/train flag correctly on accelerator model (#6877)
1c2ecbf TPUSpawn + IterableDataset error message (#6875)
87f0aea Fix DDP_SPAWN compatibility with bug_report_model.py (#6892)
eb15abc Fix finetuning complex models correctly unfreezes. (#6880)
968ac09 Remove hardcoding of rank_zero_only.rank in accelerator connector (#6878)
01b9cf8 Fix csv extension check (#6436)
86e1d9f [fix] Better support for rank_zero_only setting for SLURM and torchelastic (#6802)
b7a22ba CI: fixture for global rank variable reset (#6839)
127c52a Fix EarlyStopping logic when min_epochs not met (#6705)
last sha scrutinized: e891ceb
Including non-bugfix commits is allowed if they are necessary to support a bugfix commit and it does not break compatibility. In that case, please add a comment for why the commit was included.
Utils
To generate a list of commits after a SHA up to current master (with their corresponding links):
git log --pretty=oneline master...<last_sha_scrutinized> | sed 's/#/https:\/\/github.com\/PyTorchLightning\/pytorch-lightning\/pull\//g'
where <last_sha_scrutinized> is the SHA of the last commit that was considered to be included in the previous release.