GitLab

* [setup] support pre-build and jit-build of cuda kernels

* polish code

* polish code

* polish code

* polish code

* polish code

* polish code

* [example] add google doc for benchmark results of GPT

* add tencet doc

* [example] gpt, shard init on all processes

* polish comments

* polish code

* [builder] update readme

This reverts commit 1f8ab6f1.

[NFC] polish colossalai/auto_parallel/tensor_shard/deprecated/op_handler/embedding_handler.py code style (#2368)

Fix a typo in train_dreambooth_colossalai.py

* [workflows]bug_repair

* [workflow]new_pr_fixing_bugs
Co-authored-by: binmakeswell <binmakeswell@gmail.com>

* [device] alpha beta profiler

* add usage

* fix variable name

accelerator.num_processes -> gpc.get_world_size(ParallelMode.DATA)

* [setup] make cuda extension build optional

* polish code

* polish code

* polish code

[NFC] polish colossalai/auto_parallel/tensor_shard/deprecated/op_handler/unary_elementwise_handler.py code style (#2326)

* [workflow] added workflow to release to pypi upon version change

* polish code

* polish code

* polish code

* [doc] update link