Support transformer

[lhs8928]

• [ ] Implement multi head attention
  • [x] implement forwarding - #1961
  • [x] implement backwarding - #1963
  • [ ] implement attention mask
    • [x] support float type mask - #1963
    • [ ] support bool type mask
  • [x] verify with unittest
• [x] Implement layer normalization
  • [x] integrate core algorithm with bn layer - #1993
  • [x] implement forwarding - #1993
  • [x] verify with unittest - #1993
• [ ] Make transformer as an application
  • [x] implement positional encoding layer - #1989
  • [x] verify encoder/decoder layer with unittest - #1996
  • [ ] train with dataset

[taos-ci]

:octocat: cibot: Thank you for posting issue #1972. The person in charge will reply soon.