Microsoft might have cracked the first steps to solving one of the biggest problems in LLMs today- their inability to model long context windows well.
A lot of LLMs flex their long context windows. However, the contexts come with an unspoken problem- they tend to “overallocate attention to irrelevant context”.
This has been a huge problem in the adoption of LLMs, especially for non-technical users. The over-attention to irrelevant contexts increases Hallucinations (made up information), reduces the ability to model more nuanced relationships and much more.
In their research paper- Differential Transformer- MS researchers say the following-
Transformer tends to overallocate attention to irrelevant context. In this work, we introduce Diff Transformer, which amplifies attention to the relevant context while canceling noise. Specifically, the differential attention mechanism calculates attention scores as the difference between two separate softmax attention maps. The subtraction cancels noise, promoting the emergence of sparse attention patterns. Experimental results on language modeling show that Diff Transformer outperforms Transformer in various settings of scaling up model size and training tokens. More intriguingly, it offers notable advantages in practical applications, such as long-context modeling, key information retrieval, hallucination mitigation, in-context learning, and reduction of activation outliers. By being less distracted by irrelevant context, Diff Transformer can mitigate hallucination in question answering and text summarization. For in-context learning, Diff Transformer not only enhances accuracy but is also more robust to order permutation, which was considered as a chronic robustness issue. The results position Diff Transformer as a highly effective and promising architecture to advance large language models.
This is very exciting work and definitely worth following in the future.It has the potential to really replace the current Foundation Models-
We conduct extensive experiments on language modeling. We scale up DIFF Transformer in terms of parameter count, training tokens, and context length. The scaling curves indicate that DIFF Transformer requires only about 65% of model size or training tokens needed by Transformer to achieve comparable language modeling performance. Moreover, DIFF Transformer outperforms Transformer in various downstream tasks. The long-sequence evaluation also shows that DIFF Transformer is highly effective in utilizing the increasing context. In addition, the experimental results demonstrate that DIFF Transformer has intriguing advantages for large language models. For example, the proposed method substantially outperforms Transformer in key information retrieval, hallucination mitigation, and in-context learning. DIFF Transformer also reduces outliers in model activations, which provides new opportunities for quantization. The findings establish DIFF Transformer as an effective and distinctive foundation architecture for large language models.
Read the paper here- https://arxiv.org/abs/2410.05258
If you liked this article and wish to share it, please refer to the following guidelines.
I put a lot of effort into creating work that is informative, useful, and independent from undue influence. If you’d like to support my writing, please consider becoming a paid subscriber to this newsletter. Doing so helps me put more effort into writing/research, reach more people, and supports my crippling chocolate milk addiction. Help me democratize the most important ideas in AI Research and Engineering to over 100K readers weekly. You can use the following for an email template.
PS- We follow a “pay what you can” model, which allows you to support within your means, and support my mission of providing high-quality technical education to everyone for less than the price of a cup of coffee. Check out this post for more details and to find a plan that works for you.
Reach out to me
Use the links below to check out my other content, learn more about tutoring, reach out to me about projects, or just to say hi.
Small Snippets about Tech, AI and Machine Learning over here
AI Newsletter- https://artificialintelligencemadesimple.substack.com/
My grandma’s favorite Tech Newsletter- https://codinginterviewsmadesimple.substack.com/
Check out my other articles on Medium. : https://rb.gy/zn1aiu
My YouTube: https://rb.gy/88iwdd
Reach out to me on LinkedIn. Let’s connect: https://rb.gy/m5ok2y
My Instagram: https://rb.gy/gmvuy9
My Twitter: https://twitter.com/Machine01776819
