@calebwritescode: MiniMax model called M2 is showing its capability as an LLM with their novel interleaved reasoning that compounds in its capabilities, and also continuing with their CISPO post training method that helps improve their post-training methodologies. As LLMs from China and especially an open source model like MiniMax M2 takes the stage, we're seeing more and more examples of how strong their innovations are showing. #minimax #agi #llm #largelanguagemodels #ai #LLMs #machinelearning #Tech #research #explained #artificialintelligence #business #technology #softwareengineer

ritsu :

kimi k2 thinking

amar sofiane :

I've seen some discussion around M2 on the local LLM subreddit and the prevailing opinion is that it is significantly worse than other open source models like GLM 4.5 or Qwen 3. I think there is a big problem with how models actually perform in real life use cases vs benchmarks

Idiotsandwich :

How many AI researchers are there in China to have so many teams making LLMs

Andy Learns AI :

Always high quality stuff! What do you use to make it?

StableProjectorz :

fascinating things happening!

jaywithlime :

Great info!

Samurai Pizza Cat :

This is what I did to train my model. Adjusted the sampling weights.

Orel :

Do you have like YouTube channel I want to see in YouTube when I eat

MrBeck :

I LOVE YOUR VIDEOS you put a lot of effort into this 🔥💯

h2so4 :

how u made the video like which software

sathvik :

Huge for agentic workflows with a lot of tools

jeremynagahama :

Good info keep it up king

zacharyvmm :

Wael Rd :

❤️

Chris :

😭😭😭