Use this category to ask any question related to the course or organize study groups.
9 Likes
Hey everyone, can we form a study group similar to how MLT does it?
I really suck at keeping up with a course alone and would love to set up a weekly meeting kind of thing where we all can actually go through some parts of the course on our own and then discuss the same. We can then leave the entire week to practice it out and then discuss it briefly at the start of the next week session before repeating.
3 Likes
Sure, I’m up for it. Shall we setup a discord server?
1 Like
Yes Vinay, that’s the best way, I just set one up. Here’s a link to it: https://discord.gg/AEz4reDd .We can decide the days and time based on the people who join
1 Like
Thanks, I already joined.
1 Like
I’ve requested the kind people at MLT to help as well.
Once we find enough interest-we’ll setup the meet-ups 
7 Likes
First of all: many many thanks for putting this course online, it is a big help for me! I already worked through everything and now I can’t wait for the next part.
When do you plan to put online chapters 5 - 8?
Thank you very much!
Kiran
1 Like
hey @kiranmusze - great to hear that you enjoyed the first part and congrats on finishing it so quickly! we plan to release the next part later this year, hopefully around autumn 
I am unable to join the discord server. It seems like the link has expired.
Could you resend the invitation link? I am interested in joining the study group. Thanks!
Where can I get slides of video? Thanks!
hey @Givan we don’t plan to release the slides of the videos but almost everything is already reproduced in the course material / notebooks!
Hi Huggingface,
First of all, thank you for the excellent course! I just wanted to check if there are any new updates regarding the expected next chapters? I am really looking forward to it!
Best regards,
Jovan
1 Like
Just want to give a shoutout: many thanks and lots of love for putting this course online, it is a big help to understand the library and use it effectively.
Hopefully chapters 5 - 8 will come out soon!
Thank you very much again!
Sosig
2 Likes
Hey @Jovan we’re working on the next chapters right now and are aiming for a release sometime before the end of the year 
1 Like
Hey @lewtun,
Excellent, thanks for the reply, and special thanks for your efforts!
1 Like
I am looking for a learning path to become a huggingface 
developer. From zero to hero 
taking into account best practices. Does something like this exist? Thank you. 
@sgugger @patrickvonplaten @lysandre
Please do cover PEFT and related stuff like bitsandbytes and combining multiple adapters in detail within one of the upcoming course sections! The sooner the better. I’m sure I am not the only one who really needs this. PEFT is the unlock for finetuning SOTA models (LLaMA, Falcon, and what is yet to come) on consumer graphics cards. @philschmid, @sgugger
Hey! I am not able understand somethings . i was trying to make a chatbot which can generate reply from my custom dataset, first i trained gpt2 on my custom dataset for text generation and then i tried to finetune that on squad 2 dataset to make it work as question answering chatbot. Please help me with the approach whether i should continue down this path or any other suggestion will be good.
Hi,
I am checking the hugging face course. Its a very nice course to see how to adopt your pretrained models like BERT and GPT2/3 to develop your projects. As much as I am thankful for everyone at Hugging Face for taking this initiative so seriously, I am a little disappointed in how the material has been presented. Its like you read chap1, chap2, chap3, chap 4. chap 5 and then in chap 7 we will do integration of everything we have read so far. However, organization from the perspective of a reader should just be the opposite of this: We ought to be having chapter7 as chapter 1 or chapter 0. Like you can solve a problem bottom up manner but when presenting the solution you ought to follow Top down order i.e.,
you have to present how to use Bert and GPT3 or other LLMs in your project. First you want to cut it Bert NN before Logits and then you want to put a new header to it. And, then you want to do transfer learning with your own data for the problem you have. Like a tree traversal from the root node (starting point).
I am sorry that I have blabbered that much but without seeing the big picture first seeing all those details after details in a huge essay like manner kind of causes understanding issues. Reasoning should be presented first up for everything no matter what level you are in the tree.