[BUG] <title> deepspeed expected the next 1 parameters in the parameter fetch queue to be

[HalcyonLiang]

是否已有关于该错误的issue或讨论？ | Is there an existing issue / discussion for this?

• [X] 我已经搜索过已有的issues和讨论 | I have searched the existing issues / discussions

该问题是否在FAQ中有解答？ | Is there an existing answer for this in FAQ?

• [X] 我已经搜索过FAQ | I have searched FAQ

当前行为 | Current Behavior

RuntimeError: tracing error at step 603: module id: 1, training: True expected the next 1 parameters in the parameter fetch queue to be ({'id': 'name=transformer.visual.conv1.weight id=194', 'status': 'AVAILABLE', 'numel': 752640, 'ds_numel': 752640, 'shape': (1280, 3, 14, 14), 'ds_shape': (1280, 3, 14, 14), 'requires_grad': True, 'grad_shape': None, 'persist': False, 'active_sub_modules': {1}, 'ds_tensor.shape': torch.Size([47040])},) but got ({'id': 'name=transformer.h.2.attn.c_attn.bias id=19', 'status': 'AVAILABLE', 'numel': 6144, 'ds_numel': 6144, 'shape': (6144,), 'ds_shape': (6144,), 'requires_grad': True, 'grad_shape': None, 'persist': True, 'active_sub_modules': set(), 'ds_tensor.shape': torch.Size([384])},).

混合文本和图片数据一起训练的时候，deepzero3会有这种问题，验证了下跟数据关系不大，有大佬遇到过相同问题吗？ 看其他的issue认为是loss溢出？

期望行为 | Expected Behavior

No response

复现方法 | Steps To Reproduce

No response

运行环境 | Environment

- OS: 
- Python: 3.8
- Transformers: 4.32.0
- PyTorch: 2.0.1
- CUDA (`python -c 'import torch; print(torch.version.cuda)'`): 11.7

备注 | Anything else?

No response