google

/

flan-ul2

like 550

Hello,

do I need conversion script even if I ask transformers to download it for me?

I tried the example CPU code verbatim and got this:

Traceback (most recent call last):
  File "/Users/test/tmp/models/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 248, in __getattr__
    return self.data[item]
KeyError: 'shape'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/test/tmp/models/flan-ul5-test.py", line 10, in <module>
    outputs = model.generate(inputs, max_length=200)
  File "/Users/test/tmp/models/lib/python3.10/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "/Users/test/tmp/models/lib/python3.10/site-packages/transformers/generation/utils.py", line 1239, in generate
    batch_size = inputs_tensor.shape[0]
  File "/Users/test/tmp/models/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 250, in __getattr__
    raise AttributeError
AttributeError