AI Papers Deep Dive: Mistral 7B, ShearedLLaMA, Flash-decoding, Hypotheses-to-Theories, and more скачать в хорошем качестве

AI Papers Deep Dive: Mistral 7B, ShearedLLaMA, Flash-decoding, Hypotheses-to-Theories, and more

🚀🔍 AI papers deep dive: Mistral 7B: Explore grouped-query attention and sliding window attention techniques for enhanced performance and efficiency. https://arxiv.org/abs/2310.06825 ShearedLLaMA: How to build smaller LLMs from pruning larger models. https://xiamengzhou.github.io/sheared... Flash-decoding: Speed up long-context LLM inference by up to 8x. Hypotheses-to-Theories: Uncover how LLMs can learn rules. https://arxiv.org/abs/2310.07064 LLaVA: Very impressive open-source multimodal models https://arxiv.org/pdf/2310.03744.pdf A Review: How do large language models capture the ever-changing world knowledge? https://arxiv.org/abs/2310.07343v1 PB-LLM: extreme low-bit quantization https://arxiv.org/abs/2310.00034 OpenWebMath: A massive dataset containing every math document from the internet https://arxiv.org/abs/2310.06786 00:00 intro 00:28 Mistral 7B 03:49 ShearedLLaMA 06:57 Flash-Decoding 09:13 LLMs can learn rules 11:44 other papers 🔔 SUBSCRIBE to my channel: https://www.youtube.com/c/SophiaYangD... ⭐ Stay in touch ⭐ 📚 DS/ML Book Club: http://dsbookclub.github.io/ ▶ YouTube:    / sophiayangds   ✍️ Medium:   / sophiamyang   🐦 Twitter:   / sophiamyang   🤝 Linkedin:   / sophiamyang   💚 #ai