This is exactly what I need! One question before I dive into this awesome tutorial. This video came out on the same day as the new openai api, the assistant API -- with this new api, what would you do differently? I know that you have a two videos on it but I was wondering with this example here specifically. I think if I understand how you would use the new api in this example I would more deeply understand what the new api can do.
Hey! Thanks for the comment :) I think this tutorial is more suited for developers who are interested in training/fine-tuning a model for a bespoke purpose; like training a model to speak in a certain tone of voice, for example. You could also use this guide to train and use your model entirely offline using your own hardware, without any need to connect to an external API. The OpenAI Assistant API offers a very powerful general-purpose LLM without any need for training or fine-tuning; however, cost is a bit more of a factor. So pros and cons for both really - very much depends on your plans for the model. Hope that helps!
Running through the exact same steps you did (Wikipedia movie plots from Kaggle, limiting dataset to 1000 entries), it looks like the training is going to take about 1 hour and 45 minutes. I'm actually not that well versed in the hardware side of computers, but I can see my CPU is maxed out using an AMD Ryzen 7 3700x 8-core processor. Would this be the limiting factor causing the 7x time increase? Great tutorial, btw!
Amazing Video. This is exactly what I needed to get started. The content feels like a 10k subscriber channel and I think you will get there soon.
Thank you! So happy it helped!
Super tutorial! Keep going -you building special for anyone new to AI - code description and pace of tutorial are excellent!
Thanks so much! Means a lot!
Wow!! Amazing content! Thanks for creating it
So glad it helped!
thank you for this great video!
Super glad it was helpful! :)
This is exactly what I need! One question before I dive into this awesome tutorial. This video came out on the same day as the new openai api, the assistant API -- with this new api, what would you do differently? I know that you have a two videos on it but I was wondering with this example here specifically. I think if I understand how you would use the new api in this example I would more deeply understand what the new api can do.
Hey! Thanks for the comment :) I think this tutorial is more suited for developers who are interested in training/fine-tuning a model for a bespoke purpose; like training a model to speak in a certain tone of voice, for example. You could also use this guide to train and use your model entirely offline using your own hardware, without any need to connect to an external API. The OpenAI Assistant API offers a very powerful general-purpose LLM without any need for training or fine-tuning; however, cost is a bit more of a factor. So pros and cons for both really - very much depends on your plans for the model. Hope that helps!
Running through the exact same steps you did (Wikipedia movie plots from Kaggle, limiting dataset to 1000 entries), it looks like the training is going to take about 1 hour and 45 minutes. I'm actually not that well versed in the hardware side of computers, but I can see my CPU is maxed out using an AMD Ryzen 7 3700x 8-core processor. Would this be the limiting factor causing the 7x time increase?
Great tutorial, btw!
Hi Thanks for the video Can you please share your pc/Mac spacification I just want to compare My MacBook Air M2 training took 30 Min
Also running on an MBP M2; took around 15 minutes for me. Training time will vary slightly based on the dataset!
It's a great video. Thank you. It will be helpful if you could also attach the source code.
Working on getting stuff up on GitHub - thanks for the feedback!