Loading video...

Video Failed to Load

Go Home

How to train a reasoning model, specifically Magistral from Mistral AI? Here is a quick video walkthrough of the paper.

32,062 views • 1 year ago •via X (Twitter)

11 Comments

Sophia Yang, Ph.D.'s profile picture
Sophia Yang, Ph.D.1 year ago

Check out the full paper here:

Rainmaker's profile picture
Rainmaker2 years ago

Which Machine Learning model can beat the market? Check out this post on my free Substack where I share code and commentary for several strategies that beat passively holding a tech stock.

RobKai 🇩🇪's profile picture
RobKai 🇩🇪1 year ago

@MistralAI Thank you for this free and amazing content. How many GPUs did you use approximately?

David Webster's profile picture
David Webster1 year ago

@MistralAI thanks love the walkthrough

Jitendra Kumar Sharma's profile picture
Jitendra Kumar Sharma1 year ago

@MistralAI @sophiamyang Thanks for sharing. 🙏🙏

Kim Noël ⚡ 📖's profile picture
Kim Noël ⚡ 📖1 year ago

@MistralAI Nice thank you. But do you ever rest after the vivatech rush week ^^

⟁ndrew V's profile picture
⟁ndrew V1 year ago

@MistralAI Very helpful, thank you, and I've missed watching your videos Sophia!

Cheeku Mani Tripathi's profile picture
Cheeku Mani Tripathi1 year ago

@MistralAI Magistral is a fascinating shift RLHF without the H. Pure RL via GRPO from scratch no distillation no pretraining on reasoning traces and yet it outperforms most models trained the usual way. shows how far reward shaping and architectural tuning can take us.

Figure's profile picture
Figure1 year ago

@MistralAI skip the walkthrough, buy the drop

✨'s profile picture
1 year ago

@MistralAI Nice

SerPepe's profile picture
SerPepe1 year ago

@MistralAI Could you make a yt channel? Videos are super helpful ngl 😭

Related Videos