stp2y

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs. Source link lol

arXiv:2407.03645v1 Announce Type: new Abstract: Continual Learning (CL) involves fine-tuning pre-trained models with new data while maintaining the performance on the pre-trained data. This is particularly relevant for expanding multilingual ASR (MASR) capabilities. However, existing CL methods, mainly designed for computer vision and reinforcement learning tasks, often yield sub-optimal results when directly applied to MASR. We hypothesise that this is because CL of the auto-regressive decoder in the MASR model is difficult. To verify this, we propose four optimizations on the decoder. They include decoder-layer gradient surgery, freezing unused token embeddings, suppressing output of newly added tokens, and learning rate…

AI doesn't always produce optimal outputs. We're seeing it all the time. And contrary to popular belief, it's not because the AI system is flawed. The issue is that AI is now accessible to everyone and ChatGPT alone has 180 million active users. Create a ChatGPT account, and you can access a system that will pretty much tell you anything you need it to…but it doesn't always make sense. Creating the desired outputs is a science and an art. Dzone writer and developer evangelist Pavan Belagatti provided an insightful overview of prompt engineering in this article. As he points out, there are…