HF RL Explorer

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-oss: Agentic RL Environments (MiMo RL release). I'd like to train/run TFOpenAIGPTModel (and TFOpenAIGPTForSequenceClassification) with mixed precision and/or XLA, both of which work fine on most other TF models…

The task

I'd like to train/run TFOpenAIGPTModel (and TFOpenAIGPTForSequenceClassification) with mixed precision and/or XLA, both of which work fine on most other TF models in this repo. With OpenAI GPT specifically, neither works. AMP (mixed precision) Minimal repro: Other TF models in…

Part of XiaomiMiMo/MiMo-V2.6-RL-oss.