hummingbird2030

[Usage]: Problem when loading my trained model.

3

### Your current environment ```text Collecting environment information... PyTorch version: 2.1.2+cu121 Is debug build: False CUDA used to build PyTorch: 12.1 ROCM used to build PyTorch: N/A OS: Ubuntu 22.04.3...

Hello, I'm following your work and trying to run your code. When I install the environment in my anaconda env with `pip install -e transformers-4.28.1` I meet the following problem:...

Hello, I'm trying to generate data after training critic model using https://github.com/AkariAsai/self-rag/tree/main/data_creation/generator. But your original input data without reflection tokens seems unavailable. If you can provide original data, it will...

### Reminder - [x] I have read the above rules and searched the existing issues. ### System Info 你好。我在使用本仓库对Qwen3-8b模型全参数预训练，设备显存为8*64，使用了deepspeed zero2。 在进程中止前，grad_norm突然增大，loss比较稳定。由于报错信息较少，我无法定位问题原因。怀疑是OOM或者数据质量问题。求助如何解决？附grad_norm和loss  ### Reproduction ``` ### model model_name_or_path: /home/MODELS/Qwen/Qwen3-8B #...