Hugging Face's logo

Join the conversation

Join the community of Machine Learners and AI enthusiasts.

Sign Up
Back to feed
etemiz 
posted an update 7 days ago
Post
1725
-= DeepSeek V3 =-

After installing the new CUDA toolkit and compiling llama.cpp again I tested DeepSeek V3 yesterday.

In terms of human alignment DeepSeek V3 did worse on:
- health
- fasting
- nostr
- misinfo
- nutrition

did better on:
- faith
- bitcoin
- alternative medicine
- ancient wisdom

compared to DeepSeek 2.5. In my opinion overall it is worse than 2.5. And 2.5 wasn't that great.

There is a general tendency of models getting smarter but at the same time getting less wiser, less human aligned, less beneficial to humans.

I don't know what is causing this. But maybe synthetic dataset use for further training the LLMs makes it more and more detached from humanity. This is not going in the right direction.

My solution is to come up with a curator council to determine the datasets that are closest to human preference. "Humans that care about other humans the most" could be a definition of this dataset. What do you think?
agentlans

I have many complicated opinions about that. Not to get into a debate but I think:

  • AIs are black boxes. It's hard to say whether a new black box is really better than your old black box.
  • Even if you have received the most profound wisdom and data from the prophets - you're still training a black box.
  • AIs are more aligned with their creators than their users.
  • Any technology can be abused, no matter how well-intentioned their inventors were.
·
etemiz

What do you mean?
Everybody is also a black box until you start to talk to them. Then their ideas come out and you understand what kind of a person he/she is. I think most benchmarks are done talking to the LLMs?
Yes I am trying to use this tech in a better way, serving more humans.

In this post