في مهام توليد اللغة الطبيعية، يتم استخدام نموذج لغة عصبي لتوليد سلسلة من الكلمات التي تشكل جملة.يمكن اعتبار مصفوفة الوزن الأعلى من طراز اللغة، المعروف باسم طبقة التصنيف، كمجموعة من المتجهات، كل منها يمثل كلمة مستهدفة من قاموس الهدف.يتم تعلم ومكافحة الكلمات المستهدفة، إلى جانب بقية المعلمات النموذجية، أثناء التدريب.في هذه الورقة، نقوم بتحليل الممتلكات المشفرة في المتجهات المستهدفة والسؤال على ضرورة تعلم هذه المتجهات.نقترح تعيين ناقلات المستهدفة بشكل عشوائي وتحديدها على أنها ثابتة حتى يتم إجراء تحديثات للأوزان أثناء التدريب.نظهر أنه من خلال استبعاد ناقلات التحسين، ينخفض عدد المعلمات بشكل كبير مع تأثير هامشي على الأداء.نوضح فعالية طريقتنا في التسمية التوضيحية للصورة والترجمة الآلية.
In natural language generation tasks, a neural language model is used for generating a sequence of words forming a sentence. The topmost weight matrix of the language model, known as the classification layer, can be viewed as a set of vectors, each representing a target word from the target dictionary. The target word vectors, along with the rest of the model parameters, are learned and updated during training. In this paper, we analyze the properties encoded in the target vectors and question the necessity of learning these vectors. We suggest to randomly draw the target vectors and set them as fixed so that no weights updates are being made during training. We show that by excluding the vectors from the optimization, the number of parameters drastically decreases with a marginal effect on the performance. We demonstrate the effectiveness of our method in image-captioning and machine-translation.
References used
https://aclanthology.org/
We ask subjects whether they perceive as human-produced a bunch of texts, some of which are actually human-written, while others are automatically generated. We use this data to fine-tune a GPT-2 model to push it to generate more human-like texts, an
The advent of Deep Learning and the availability of large scale datasets has accelerated research on Natural Language Generation with a focus on newer tasks and better models. With such rapid progress, it is vital to assess the extent of scientific p
Knowledge-enriched text generation poses unique challenges in modeling and learning, driving active research in several core directions, ranging from integrated modeling of neural representations and symbolic information in the sequential/hierarchica
We propose an approach to automatically test for originality in generation tasks where no standard automatic measures exist. Our proposal addresses original uses of language, not necessarily original ideas. We provide an algorithm for our approach an
We introduce GEM, a living benchmark for natural language Generation (NLG), its Evaluation, and Metrics. Measuring progress in NLG relies on a constantly evolving ecosystem of automated metrics, datasets, and human evaluation standards. Due to this m