Sentiment Analysis On Imdb

المقاييس

Accuracy

النتائج

نتائج أداء النماذج المختلفة على هذا المعيار القياسي

اسم النموذج	Accuracy	Paper Title
ResNext[resnext]	85	Classifying Textual Data with Pre-trained Vision Models through Transfer Learning and Data Transformations
RoBERTa.base	95.79	The Document Vectors Using Cosine Similarity Revisited
ToWE-SG	90.8	Task-oriented Word Embedding for Text Classification
Nyströmformer	93.2	Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention
FLAN 137B (zero-shot)	94.3	Finetuned Language Models Are Zero-Shot Learners
Doc2VecC	88.3	Efficient Vector Representation for Documents through Corruption
DV-ngrams-cosine + NB-weighted BON (re-evaluated)	93.68	The Document Vectors Using Cosine Similarity Revisited
BERT large finetune UDA	95.8	Unsupervised Data Augmentation for Consistency Training
Llama-2-70b-chat (0-shot)	95.39	LlamBERT: Large-scale low-cost data annotation in NLP
XLNet	96.21	XLNet: Generalized Autoregressive Pretraining for Language Understanding
GraphStar	96.0	Graph Star Net for Generalized Multi-Task Learning
BCN+Char+CoVe	91.8	Learned in Translation: Contextualized Word Vectors
OCaTS (kNN & GPT-3.5-turbo	93.06	Cache me if you Can: an Online Cost-aware Teacher-Student framework to Reduce the Calls to Large Language Models
coRNN	87.4%	Coupled Oscillatory Recurrent Neural Network (coRNN): An accurate and (gradient) stable architecture for learning long time dependencies
FLAN 137B (few-shot, k=2)	95	Finetuned Language Models Are Zero-Shot Learners
DV-ngrams-cosine	93.13	Sentiment Classification Using Document Embeddings Trained with Cosine Similarity
CNN+LSTM	88.9	On the Role of Text Preprocessing in Neural Network Architectures: An Evaluation Study on Text Categorization and Sentiment Analysis
BERT_base+ITPT	95.63	How to Fine-Tune BERT for Text Classification?
DV-ngrams-cosine + RoBERTa.base	95.92	The Document Vectors Using Cosine Similarity Revisited
DV-ngrams-cosine with NB sub-sampling + RoBERTa.base	95.94	The Document Vectors Using Cosine Similarity Revisited

0 of 48 row(s) selected.