t5 conversion example error

[affereaflaw]

Description of the bug:

I am quite new in this field, i want to try t5 conversion using the colab version (https://github.com/google-ai-edge/ai-edge-torch/blob/main/ai_edge_torch/generative/examples/t5/t5_conversion_colab.ipynb)

I got this error when running the last cell

converting T5 to tflite.

TypeError Traceback (most recent call last)
in <cell line: 0>()
1 print('converting T5 to tflite.')
----> 2 edge_model = convert_t5_to_tflite_multisig("/content/pytorch_model.bin")

6 frames
/usr/local/lib/python3.11/dist-packages/ai_edge_torch/generative/examples/t5/t5_attention.py in init(self, batch, dim, config, norm_config, kv_cache_max, enable_hlfb, has_relative_attention_bias)
156 has_relative_attention_bias (bool): whether we compute relative bias.
157 """
--> 158 super().init(batch, dim, dim, config, kv_cache_max, enable_hlfb)
159 self.pre_atten_norm = builder.build_norm(dim, norm_config)
160

TypeError: CrossAttention.init() missing 1 required positional argument: 'enable_hlfb'

Thank you

Actual vs expected behavior:

Any other information you'd like to share?

[gaikwadrahul8]

Hi, @affereaflaw
Thank you for bringing this issue to our attention. I have been able to reproduce the same behavior on my end and am receiving the same error message: TypeError: CrossAttention.__init__() missing 1 required positional argument: 'enable_hlfb'. For reference, I have attached gist-file demonstrating the issue. We will investigate this further and provide an update as soon as possible.