文件列表:

.codecov.yml (260, 2023-01-20)
.coveragerc (60, 2023-01-20)
.dockerignore (54, 2023-01-20)
.dvc/ (0, 2023-01-20)
.dvc/config (142, 2023-01-20)
.dvcignore (139, 2023-01-20)
.env_template (47, 2023-01-20)
.pre-commit-config.yaml (774, 2023-01-20)
LICENSE (1079, 2023-01-20)
Makefile (5346, 2023-01-20)
cloudbuild.yml (678, 2023-01-20)
configs/ (0, 2023-01-20)
configs/debug_cpu.yaml (153, 2023-01-20)
configs/train_cpu.yaml (159, 2023-01-20)
configs/train_gpu.yaml (158, 2023-01-20)
data/ (0, 2023-01-20)
data/external/ (0, 2023-01-20)
data/interim/ (0, 2023-01-20)
data/processed/ (0, 2023-01-20)
data/raw/ (0, 2023-01-20)
data/raw/Fake.csv.dvc (80, 2023-01-20)
data/raw/True.csv.dvc (80, 2023-01-20)
docs/ (0, 2023-01-20)
docs/Makefile (5620, 2023-01-20)
docs/commands.rst (489, 2023-01-20)
docs/conf.py (8021, 2023-01-20)
docs/getting-started.rst (256, 2023-01-20)
docs/index.rst (453, 2023-01-20)
docs/make.bat (5124, 2023-01-20)
models/ (0, 2023-01-20)
... ...

[](https://github.com/RSM-git/mlops_fake_real_news/actions/workflows/tests.yml) [](https://codecov.io/gh/RSM-git/mlops_fake_real_news) ## Machine Learning Operations project description - NLP with fake and real news articles Casper Brun Pedersen - s204119
Marcus Presutti - s204122
Rasmus Steen Mikkelsen - s204135
Victor Tolsager Olesen - s204141 ### Overall goal of the project The goal of the project is to develop an MLOps pipeline for classifying news articles as either fake or real using Transformers. ### What framework are you going to use (PyTorch Image Models, Transformers, PyTorch-Geometrics) The model will be trained on text data, which prompts the use of the Transformers framework. ### How do you intend to include the framework in your project? The transformers framework contains a multitude of pretrained models. We intend on fine-tuning the pretrained ALBERT transformer ### What data are you going to run on? (initially, may change) We will be using a Kaggle dataset, specifically the [Fake and real news dataset](https://www.kaggle.com/datasets/clmentbisaillon/fake-and-real-news-dataset). The dataset consists of two `.csv` files `Fake.csv` and `True.csv`, which will be merged into a single dataset. Each sample initially has four attributes: `title`, `text`, `subject`, and `date`, and will naturally get a label attribute based on which file the sample was initially in. ### What deep learning models do you expect to use? We have intentions of using the pretrained [ALBERT](https://huggingface.co/docs/transformers/model_doc/albert) from huggingface, which will be fine-tuned using the fake and real news dataset. ALBERT performs parameter reduction techniques to lower memory consumption and increase training speed of the original [BERT](https://huggingface.co/docs/transformers/model_doc/bert) model.

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：