birgermoell/swedish-gpt - Каталог нейросетей
Генерация текста

birgermoell/swedish-gpt

Добавлено:
birgermoell/swedish-gpt

This model is part of a series of models training on TPU with Flax Jax during Huggingface Flax/Jax challenge. https://huggingface.co/flax-community/swe-roberta-wiki-oscar A swedish GPT2 style model trained using Flax CLM pipeline on the Swedish part of the wiki40b dataset and the Oscar dataset. https://huggingface.co/datasets/wiki40b The model was trained for around 22600 steps (42 hours) as part of the Huggingface Jax/Flax challenge with the following loss and learning rate Loss: 3.1715331077575684, Learning Rate: 0.0024816440418362617) The data was cleaned and preprocessed using the following script. Make sure to install depencies for beam_runner to make the dataset work.

Модальности:
Генерация текста


Задача: Генерация текста
Автор: birgermoell
Теги: jax, tensorboard, gpt2, sv, text-generation-inference, endpoints_compatible
Лайков: 2  |  Загрузок: 157

Открыть на HuggingFace →

Описание основано на материалах HuggingFace. Перевод выполнен автоматически.