Finn

update gradio version, on version 3.29.0 the bug is fixed

> https://github.com/microsoft/DeepSpeed/blob/4de4d2acc6077707c806a8abbc26fba1e2779745/deepspeed/ops/transformer/inference/ds_attention.py#L101 > > 1 - input_mask need to be modified to ~input_mask Solved my problem, thanks

你好，能分享我一份天池NL2SQL的数据集吗？万分感谢[email protected]

> > > > 是在data_utils.py中，在train_phase=3下修改吗？比如 `tokenizer(prompt, return_tensors="pt", max_length=max_seq_len, padding="max_length", truncation=True)` 我这边的运行日志中有一些act loss和cti loss是inf或者非常大的数值，然后就会报 Current loss scale already at minimum - cannot decrease scale anymore. Exiting run. 这个错误 `epoch: 0|step: 46|ppo_ep:...

I get the same error , torch 1.13+cu117

> I have built a docker image for it `docker pull zhyict/verl:verl0.7.0-transformers4.57.1-vllm0.11.0-mcore0.13.1` **which cudnn version is used? I met the error below. Other dependency versions are consistent with yours** File...

> Here is a step-by-step guide to implement the solution: > > **Step 1: Override the Default Reward Function** > > Modify the file at `/root/PycharmProjects/pythonProject/.think_with_img/verl/verl/utils/reward_score/__init__.py` by overriding the `default_compute_score`...

这个问题有解决吗？我也是在保存大模型的时候有问题，小点的模型没问题

save model的时候报这个错，模型比较大，gpt-oss-120B，保存一半就挂了，有解决方法么？@hiyouga