mikeshev4enko's profile picture. Founder D-PC Messenger. Building open-source P2P network for sharing AI resources and creating human-AI collective intelligence. Seeking a business co-founder.

Mike Shevchenko

@mikeshev4enko

Founder D-PC Messenger. Building open-source P2P network for sharing AI resources and creating human-AI collective intelligence. Seeking a business co-founder.

Privacy or god-tier knowledge from AI? You can only keep one when collaborating. Choose


Mike Shevchenko reposted

@miromind_ai released their open source latest model on research agent. - interleaved thinking with multi-step analysis, powered by RL - 256k context window - handles up to 600 tool calls per task - comes in 8B, 30B, and 72B for different compute budgets - MiT license Model…

Xianbao_QIAN's tweet image. @miromind_ai released their open source latest model on research agent.

- interleaved thinking with multi-step analysis, powered by RL
- 256k context window
- handles up to 600 tool calls per task
- comes in 8B, 30B, and 72B for different compute budgets
- MiT license

Model…

Mike Shevchenko reposted

Excited to release new repo: nanochat! (it's among the most unhinged I've written). Unlike my earlier similar repo nanoGPT which only covered pretraining, nanochat is a minimal, from scratch, full-stack training/inference pipeline of a simple ChatGPT clone in a single,…

karpathy's tweet image. Excited to release new repo: nanochat!
(it's among the most unhinged I've written).

Unlike my earlier similar repo nanoGPT which only covered pretraining, nanochat is a minimal, from scratch, full-stack training/inference pipeline of a simple ChatGPT clone in a single,…

Moon rover drift. Made by @grok Imagine


Mike Shevchenko reposted

I had an interesting experience with Claude Sonnet 4. When it made mistakes, I asked it to conduct a deep and detailed self-reflection. After it described its thoughts about the mistake and provided mitigations, I asked it to create a JSON file with improved instructions for…

mikeshev4enko's tweet image. I had an interesting experience with Claude Sonnet 4. When it made mistakes, I asked it to conduct a deep and detailed self-reflection. After it described its thoughts about the mistake and provided mitigations, I asked it to create a JSON file with improved instructions for…

Who wants to be a billionaire? Just create a Cursor for non-programming tasks using simple open source technology github.com/mikhashev/pers…


Mike Shevchenko reposted

The prompt arms race between users and model providers begins!


Errors are information about abstraction boundaries.


I think about key principles for AI-enhanced learning, what missed ?

mikeshev4enko's tweet image. I think about key principles for AI-enhanced learning, what missed ?

United States Trends

Loading...

Something went wrong.


Something went wrong.