使用p-tuning去finetune glm-large-chinese模型时 --continuous-prompt

[Chenchenwei]

File "/opt/conda/lib/python3.8/site-packages/deepspeed/init.py", line 125, in initialize
engine = DeepSpeedEngine(args=args,
File "/opt/conda/lib/python3.8/site-packages/deepspeed/runtime/engine.py", line 340, in init
self._configure_optimizer(optimizer, model_parameters)
File "/opt/conda/lib/python3.8/site-packages/deepspeed/runtime/engine.py", line 1291, in _configure_optimizer
self.optimizer = self._configure_zero_optimizer(basic_optimizer)
File "/opt/conda/lib/python3.8/site-packages/deepspeed/runtime/engine.py", line 1542, in _configure_zero_optimizer
optimizer = DeepSpeedZeroOptimizer(
File "/opt/conda/lib/python3.8/site-packages/deepspeed/runtime/zero/stage_1_and_2.py", line 527, in init
self._link_all_hp_params()
File "/opt/conda/lib/python3.8/site-packages/deepspeed/runtime/zero/stage_1_and_2.py", line 555, in _link_all_hp_params
link_hp_params(
File "/opt/conda/lib/python3.8/site-packages/deepspeed/utils/mixed_precision_linkage.py", line 20, in link_hp_params
lp_param._hp_mapping = get_hp_fragment_mapping(lp_param,
File "/opt/conda/lib/python3.8/site-packages/deepspeed/utils/tensor_fragment.py", line 76, in get_hp_fragment_mapping
assert fragment_start < fragment_end,
AssertionError: fragment start 355338240 should be < fragment_end 355338240

[Xuanrui Lin]

你好，我也遇到了相同的问题，请问你是怎么解决的呀

[mechigonft]

你好，我也遇到了同样的问题，请问你怎么解决的？万分感谢

[980202006]

There is a high probability that the shape of a certain parameter is wrong, such as torch.randn(0).