TF OpenAI GPT model fails under mixed precision (AMP) and XLA

TF OpenAI GPT model fails under mixed precision (AMP) and XLA: a task in MiMo-V2.6-RL-harbor-code: MiMo-V2.6-RL Code (Harbor) (Harbor dataset). I'd like to train/run TFOpenAIGPTModel (and TFOpenAIGPTForSequenceClassification ) with mixed precision and/or XLA, both of which work fine on most other…