DeepSeek released their new flagship model V4 which yet again shocks the world but not in training cost but this time in inference cost by reducing compute and inference demand by a huge margin.
As we continue our AI race between open labs and closed labs, V4 sets another leap into how LLMs are becoming more lean and intelligent since two nations have different incentive in how they should grow.
#deepseek #ai #llm
Sign up for Intuive AI (ByCloud):
https://www.intuitiveai.academy/
40% OFF Use Coupon Code: CALEB
Chapters
00:00 Intro
00:45 Flagship Model
01:32 Open vs Closed
02:23 Cost of Intelligence
03:24 Paper
04:35 Sponsor: ByCloud
05:15 Pricing
06:18 Attention
07:16 DSA
08:23 CSA, HCA
10:19 Conclusion
