This paper presents a neural attention-based model for abstractive sentence summarization, which generates summaries by conditioning each word on the input sentence. The model combines a neural language model with a contextual input encoder, inspired by recent advancements in neural machine translation. The encoder learns a soft alignment between the input text and the generated summary, allowing the model to incorporate the input text into the generation process. The model is trained end-to-end on a large dataset of article-summary pairs and shows significant performance improvements over several strong baselines in the DUC-2004 shared task. The paper also discusses various encoders, including bag-of-words, convolutional, and attention-based encoders, and introduces a modified scoring function to balance abstractive and extractive tendencies. Experimental results demonstrate the effectiveness of the proposed model, outperforming both machine translation systems and traditional summarization methods.This paper presents a neural attention-based model for abstractive sentence summarization, which generates summaries by conditioning each word on the input sentence. The model combines a neural language model with a contextual input encoder, inspired by recent advancements in neural machine translation. The encoder learns a soft alignment between the input text and the generated summary, allowing the model to incorporate the input text into the generation process. The model is trained end-to-end on a large dataset of article-summary pairs and shows significant performance improvements over several strong baselines in the DUC-2004 shared task. The paper also discusses various encoders, including bag-of-words, convolutional, and attention-based encoders, and introduces a modified scoring function to balance abstractive and extractive tendencies. Experimental results demonstrate the effectiveness of the proposed model, outperforming both machine translation systems and traditional summarization methods.