Add some helpers: - [x] Literal: add a helper to create F32 / U32 literals and use in the GPT2 implementation - [ ] LayerNorm - [x] Move LayerNorm to stdlib, reduce boilerplate using helpers - [ ] Create a `Module` for `LayerNorm` operating on rank-2 tensors, and a helper to call it - [ ] Squeeze / Unsqueeze helpers
Add some helpers:
ModuleforLayerNormoperating on rank-2 tensors, and a helper to call it