Train Your Own Reasoning Model Like DeepSeek R1 in Free Google Colab

This video is a step-by-step tutorial to train your own reasoning model like R1 with GRPO in google colab with Unsloth.
🔥 Get 50% Discount on any A6000 or A5000 GPU rental, use following link and coupon:
bit.ly/fahd-mirza
Coupon code: FahdMirza
🔥 Buy Me a Coffee to support the channel: ko-fi.com/fahd...
🚀 This video is sponsored by EigentBOT that lets you deploy a personalized knowledge bot across platforms like Discord, Slack, etc. bot.eigent.ai
▶ Become a Patron 🔥 - / fahdmirza
#deepseekr1 #tgi #unsloth
PLEASE FOLLOW ME:
▶ LinkedIn: / fahdmirza
▶ KZbin: / @fahdmirza
▶ Blog: www.fahdmirza.com
RELATED VIDEOS:
▶ Resource colab.research...
All rights reserved © Fahd Mirza

[@Faiman1000]

You old chap art one of the finest upon this videlicet platform-verily, thou deserv’st every triumph! A true gentleman and scholar, young sir, of most estimable bearing!

[@solomslls]

can you do a local implémentation of this please, and maybe change the model so we get an idea of what we need to change , please, there are lot of old good models would love to train them on reasoning, and thank you

[@fahdmirza]

same steps can be used in local jupyter notebook or in python code.

[@solomslls]

@fahdmirza I see, is it possible for Jupyter notebook to use my local llm folder directly from pc ?

[@areaadwait4657]

This didn't work for the gemma-2 models . Not sure why it keeps pulling the gemma-2-bnb-4bit models.

[@fahdmirza]

Thats not for gemma

[@datadiggas]

First!

[@fahdmirza]

cheers

[@arsalanarsalan1098]

Second