How to fine tune LLM | How to fine tune Chatgpt | How to fine tune llama3

[@shourav2368]

This is the best channel for learning cs ❤❤

[@UnfoldDataScience]

Thanks a lot - please share with friends as well.

[@paulaugustus4671]

Thanks for the amazing video.

[@KumR]

Hi Aman . Do u mean during back propagation some weights are ignored using sokething like no grad?

[@ARkhan-xw8ud]

Hmmm yes during back propagation in transformer some weights are ignored and to solve this problem they come up with learning parameters

[@UnfoldDataScience]

Are you asking about LoRa or something else that we discussed in video

[@suryaravi3907]

Why get_peft_model is not used in the trainer model parameter?

[@UnfoldDataScience]

I will check script and get back

[@AsheeshKumar-w3n]

Thank you Sir!
I really like the way you simplify these complex concepts.
For in depth future videos, I would request to go in deep which is commonly used in the industry project.❤

[@UnfoldDataScience]

Yes sure - will do

[@abkk2460]

Sir Thank you for the helpful explanation. Looking forward to more detail WRT Azure Open AI. Tnxs

[@UnfoldDataScience]

As soon as possible

[@ARkhan-xw8ud]

LORA stands for Low Rank Adaption

[@UnfoldDataScience]

Thank you.

[@madial382]

can u make full llm serice

[@UnfoldDataScience]

Many videos are coming please press bell icon to get all notifications

[@akshaymahajan7971]

I wanted to reach you sir what is the procedure for that

[@UnfoldDataScience]

You can email me. Email id in channel page.

[@DarkLord79799]

Don't use simple dataset, use atleats have 2 to 4 features like instruction, input, output, for finetunning llm it's help a lot

[@UnfoldDataScience]

Many thanks for watching, sure will do

[@soumyanath5595]

Planning to fine tune GPT 2 small LLM model (145 million parameters.)
Will a laptop consisting of i5 13th Gen and Nvidia RTX 3050 6GB Graphics card, able to do all this task

[@UnfoldDataScience]

Use small batch sizes and gradient accumulation.
Enable mixed-precision training (FP16) with libraries like PyTorch (torch.cuda.amp).

[@AsheeshKumar-w3n]

Thank you Sir!
I really like the way you simplify these complex concepts.
For in depth future videos, I would request to go in deep which is commonly used in the industry project.❤

[@UnfoldDataScience]

Thanks a lot - many videos on similar topics are on the way. Just ensure you press the bell icon