Live data from Hacker News

DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence

huggingface.co

21–23 of 23 posts

Re: DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence

#21

From this thread [0] I can assume that because, while 1.6T, it is A49B, it can run (theoretically, very slow maybe) locally on consumer hardeware, or is that wrong? [0] https://news.ycombinator.com/item?id=47864835

The flash version is smaller, I think around 200B parameters and is cheap to run.

Re: DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence

#22

From this thread [0] I can assume that because, while 1.6T, it is A49B, it can run (theoretically, very slow maybe) locally on consumer hardeware, or is that wrong? [0] https://news.ycombinator.com/item?id=47864835

No because you don't know which 49 is needed until the moment it's needed

Re: DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence

#23

From this thread [0] I can assume that because, while 1.6T, it is A49B, it can run (theoretically, very slow maybe) locally on consumer hardeware, or is that wrong? [0] https://news.ycombinator.com/item?id=47864835

It will be Seconds Per Token instead of Tokens Per Second.

2 s/t on to NVMes

The issue is KV cache

Post reply on HN