Tensorflow TPUStrategy and model.generate: Does it work?

JulesGM · May 7, 2021, 2:57pm

Hello folks,

I’m trying to get GPT-2 in Tensorflow 2.x to generate on TPUs using tf.distribute.TPUStrategy. This requires compiling the functions with tf.function, but I haven’t had any luck getting tf.function to work on a TFGPT2LMHeadModel.

Has anyone tried this, and / or had any luck?
Thanks.

Topic		Replies	Views
[Tensorflow Export] How to export a fine tuned GPT2 model to a tensorflow model file? Beginners	1	411	January 15, 2021
GPT2 with TensorFlow? 🤗Transformers	1	330	November 14, 2020
How to load finetuned model in TF Beginners	2	357	September 28, 2020
How to use generation of gpt2 from huggingface transformers in tensorflow keras model? Models	0	1038	June 8, 2021
Which version of Transformers and Tensorflow were used in the script below? 🤗Transformers	0	219	January 4, 2023

Tensorflow TPUStrategy and model.generate: Does it work?

Related Topics