حققت نماذج تمثيل اللغة المدربة مؤخرا مثل بيرت وروبرتا نتائج مهمة في مجموعة واسعة من مهام معالجة اللغة الطبيعية (NLP)، ومع ذلك، فإنها تتطلب تكلفة حسابية عالية للغاية.يعد تعلم المناهج الدراسية (CL) أحد الحلول المحتملة لتخفيف هذه المشكلة.CL هي استراتيجية تدريبية حيث يتم إعطاء عينات التدريب للنماذج بأمر هادف بدلا من أخذ العينات العشوائية.في هذا العمل، نقترح طريقة CL جديدة تدريجيا، مما يزيد تدريجيا من حجم نص المدخلات لتدريب آلية الانتباه الذاتي في بيرت ومتغيراته باستخدام الحد الأقصى للحجم الدفعة المتوفرة.تظهر التجارب في إعدادات الموارد المنخفضة أن نهجنا يفوق أن يؤدي إلى خط الأساس من حيث سرعة التقارب والأداء النهائي على مهام المصب.
Recently, pre-trained language representation models such as BERT and RoBERTa have achieved significant results in a wide range of natural language processing (NLP) tasks, however, it requires extremely high computational cost. Curriculum Learning (CL) is one of the potential solutions to alleviate this problem. CL is a training strategy where training samples are given to models in a meaningful order instead of random sampling. In this work, we propose a new CL method which gradually increases the block-size of input text for training the self-attention mechanism of BERT and its variants using the maximum available batch-size. Experiments in low-resource settings show that our approach outperforms the baseline in terms of convergence speed and final performance on downstream tasks.
References used
https://aclanthology.org/
This paper investigates and reveals the relationship between two closely related machine learning disciplines, namely Active Learning (AL) and Curriculum Learning (CL), from the lens of several novel curricula. This paper also introduces Active Curri
Recent researches show that pre-trained models (PTMs) are beneficial to Chinese Word Segmentation (CWS). However, PTMs used in previous works usually adopt language modeling as pre-training tasks, lacking task-specific prior segmentation knowledge an
We take an IDEA Algorithm and add to it some stages depend on
BBM to get an Enhanced Algorithm, which had 3keys, 128-bit
input block.
In this work, we focus on a more challenging few-shot intent detection scenario where many intents are fine-grained and semantically similar. We present a simple yet effective few-shot intent detection schema via contrastive pre-training and fine-tun
We address the compositionality challenge presented by the SCAN benchmark. Using data augmentation and a modification of the standard seq2seq architecture with attention, we achieve SOTA results on all the relevant tasks from the benchmark, showing t