Ask the llm to summarize stuff when it's running low, should be better than just rememberig the first bit and a window.     See Less

🎯 Key Takeaways for quick navigation:

00:00 📚 Increasing Data Input for Large Language Models
- Large languag     See More ace challenges with increased data input.
- GPU memory limitations and computational time impact performance.
- The concept of "Window attention" has been used to mitigate these issues.
01:09 🔄 Introducing StreamingLM and Attention Sync
- StreamingLM is a research project to enhance data input for large language models.
- Attention Sync focuses on the importance of initial tokens for context.
- StreamingLM combines initial tokens and a rolling cache for effective context.
02:49 🔓 Unlocking Possibilities with StreamingLM
- StreamingLM enables handling long-form content generation and movie transcripts.
- It works well for scenarios that require generating a large amount of content.
- However, it may not handle extremely complex tasks with extensive context loss.

Made with HARPA AI    See Less

is there is still no solution for extending data?     See Less

wait wait! you just finished too quickly I hope you have elaborated more on what can't be accomplished I didn't quite get why and the difference between what we can achieve with long     See More ents.    See Less

as usual, inspiring, accurate and updated. ty sir     See Less

Which LLM has the largest token limit to expand the context length of the chat?     See Less

This is my third video I’ve seen today from you and you are so consistent with providing value with your words. Thank you my new AI guru🙏     See Less

i luv u     See Less

I don't understand how can it help even for books. Will it forget everything in the middle of the book? I try to think how it works in human brain. When we reed a book we (usually) don&#     See More ber each word. What we do, we create visual images inside and it compress the book very effectively. Supposedly, these images are like tokens or maybe like embeddings and don't occupy much space in memory. Is it possible to implement something like this for LLMs? They should kind of learn during "reading the book" and they should convert texts to multimodal embeddings or even find (create) approximate path in embedding space and later they should have the ability to analyze this path later. Not sure how it should be implemented.    See Less

Great job at providing information about new developments, Jason! Thanks!     See Less