norsk-gpt-wiki

flax-community/norsk-gpt-wiki

norsk-gpt-wiki is a text generation model from flax-community. Use it when you need the model to write or continue text. It is set up for transformers.

A norwegian GPT2 style model trained using Flax CLM pipeline on the Norwegian part of the wiki40b dataset.

Downloads · 30 days

251

9% of all-time downloads

All-time downloads

2.9K

Public

Parameters

137M

1.5 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.bin510 MB · 34%

Parameter types

How the weights are stored.

F32124M · 91%

Try a prompt

Spaces

birgermoell/language-explorer

Model card

From the Hugging Face model README

GPT2-svenska-wikipedia

A norwegian GPT2 style model trained using Flax CLM pipeline on the Norwegian part of the wiki40b dataset.

https://huggingface.co/datasets/wiki40b

Model series

This model is part of a series of models training on TPU with Flax Jax during Huggingface Flax/Jax challenge.

Gpt models

Roberta models

Nordic Roberta Wiki

https://huggingface.co/flax-community/nordic-roberta-wiki

Swe Roberta Wiki Oscar

https://huggingface.co/flax-community/swe-roberta-wiki-oscar

Roberta Swedish Scandi

https://huggingface.co/birgermoell/roberta-swedish-scandi

Roberta Swedish

https://huggingface.co/birgermoell/roberta-swedish

Swedish T5 model

https://huggingface.co/birgermoell/t5-base-swedish

Data cleaning and preprocessing

The data was cleaned and preprocessed using the following script. Make sure to install depencies for beam_runner to make the dataset work.

from datasets import load_dataset
def load_and_clean_wiki():
    dataset = load_dataset('wiki40b', 'no', beam_runner='DirectRunner', split="train")
    #dataset = load_dataset('wiki40b', 'sv', beam_runner='DirectRunner')
    dataset = dataset.remove_columns(['wikidata_id', 'version_id'])
    filtered_dataset = dataset.map(filter_wikipedia)
    # filtered_dataset[:3]
    # print(filtered_dataset[:3])
    return filtered_dataset

def filter_wikipedia(batch):
    batch["text"] = " ".join(batch["text"].split("\
_START_SECTION_\
"))
    batch["text"] = " ".join(batch["text"].split("\
_START_ARTICLE_\
"))
    batch["text"] = " ".join(batch["text"].split("\
_START_ARTICLE_\
"))
    batch["text"] = " ".join(batch["text"].split("\
_START_PARAGRAPH_\
"))
    batch["text"] = " ".join(batch["text"].split("_NEWLINE_"))
    batch["text"] = " ".join(batch["text"].split("\xa0"))
    return batch

Training script

The following training script was used to train the model.

./run_clm_flax.py     --output_dir="${MODEL_DIR}"     --model_type="gpt2"     --config_name="${MODEL_DIR}"     --tokenizer_name="${MODEL_DIR}"     --dataset_name="wiki40b"     --dataset_config_name="no"     --do_train --do_eval     --block_size="512"     --per_device_train_batch_size="64"     --per_device_eval_batch_size="64"     --learning_rate="5e-3" --warmup_steps="1000"     --adam_beta1="0.9" --adam_beta2="0.98" --weight_decay="0.01"     --overwrite_output_dir     --num_train_epochs="20"     --logging_steps="500"     --save_steps="1000"     --eval_steps="2500"     --push_to_hub

Task: Text Generation
Library: transformers
Type: gpt2
Languages: no
Created: Mar 2, 2022
Updated: Sep 11, 2023