{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/practical-text-classification-with-large-pre","title":"Practical Text Classification With Large Pre-Trained Language Models","arxiv_id":"1812.01207","date":"2018-12-04","proceeding":null,"authors":["Neel Kant","Raul Puri","Nikolai Yakovenko","Bryan Catanzaro"],"abstract":"Multi-emotion sentiment classification is a natural language processing (NLP)\nproblem with valuable use cases on real-world data. We demonstrate that\nlarge-scale unsupervised language modeling combined with finetuning offers a\npractical solution to this task on difficult datasets, including those with\nlabel class imbalance and domain-specific context. By training an\nattention-based Transformer network (Vaswani et al. 2017) on 40GB of text\n(Amazon reviews) (McAuley et al. 2015) and fine-tuning on the training set, our\nmodel achieves a 0.69 F1 score on the SemEval Task 1:E-c multi-dimensional\nemotion classification problem (Mohammad et al. 2018), based on the Plutchik\nwheel of emotions (Plutchik 1979). These results are competitive with state of\nthe art models, including strong F1 scores on difficult (emotion) categories\nsuch as Fear (0.73), Disgust (0.77) and Anger (0.78), as well as competitive\nresults on rare categories such as Anticipation (0.42) and Surprise (0.37).\nFurthermore, we demonstrate our application on a real world text classification\ntask. We create a narrowly collected text dataset of real tweets on several\ntopics, and show that our finetuned model outperforms general purpose\ncommercially available APIs for sentiment and multidimensional emotion\nclassification on this dataset by a significant margin. We also perform a\nvariety of additional studies, investigating properties of deep learning\narchitectures, datasets and algorithms for achieving practical multidimensional\nsentiment classification. Overall, we find that unsupervised language modeling\nand finetuning is a simple framework for achieving high quality results on\nreal-world sentiment classification.","url_abs":"http://arxiv.org/abs/1812.01207v1","url_pdf":"http://arxiv.org/pdf/1812.01207v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"practical-text-classification-with-large-pre","repo_url":"https://github.com/NVIDIA/sentiment-discovery","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"emotion-classification","task_name":"Emotion Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"sentiment-analysis","task_name":"Sentiment Analysis"},{"task_slug":"sentiment-classification","task_name":"Sentiment Classification"},{"task_slug":"text-classification","task_name":"Text Classification"},{"task_slug":"text-classification-1","task_name":"text-classification"}],"methods":[{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"transformer","method_name":"Transformer"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/emotion-classification-on-semeval-2018-task","task":"Emotion Classification","dataset":"SemEval 2018 Task 1E-c","model":"Transformer (finetune)","rank_in_archive_order":4,"of":4,"metrics":{"Macro-F1":"0.561"},"uses_additional_data":false},{"leaderboard":"/sota/sentiment-analysis-on-sst-2-binary","task":"Sentiment Analysis","dataset":"SST-2 Binary classification","model":"Transformer (finetune)","rank_in_archive_order":59,"of":87,"metrics":{"Accuracy":"90.9"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1812.01207","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}