Merge pull request !2748 from ghzl/change-order-params-only-equal-to-group-param

Merge pull request !2286 from gziyan/optimizer_parallel

Merge pull request !2399 from ghzl/fix-params-keyerror-in-group-params

Merge pull request !2303 from gziyan/modify_lars_interface

- It provides a unified style to express graph and kernel for user.
- It provides a unified IR to represent graph and kernel for developer.
- It breaks the boundary between graph and kernel.
- It provides more opportunities to do compile optimization.

Merge pull request !2381 from liangzelang/dev

Merge pull request !2071 from wangnan39/support_loss_scale_for_sparse_optimizer

Merge pull request !2112 from yoonlee666/adam

Merge pull request !2110 from lilei/add_proximal_ada_grad_optimizer

Merge pull request !2007 from wangnan39/add_lazy_adam_optim_and_support_sparse_admm_for_cpu_backend

Merge pull request !1862 from jiangjinsheng/issue_doc

Merge pull request !1831 from ghzl/add-oder-parameters-in-group-functions

Merge pull request !1890 from liuxiao/fix-for-issuse

Merge pull request !1896 from gziyan/fix_lars_weight_decay

Merge pull request !1826 from wangnan39/fix_lr_check_bug_in_adamweightdecay_dynamic_lr

Merge pull request !1777 from BowenK/master

Merge pull request !1542 from vlne-v1/I1GZ0B-multitype-funcgraph-bug

Merge pull request !1555 from wangnan39/fix_bug_in_check_lamb_warmup_step