Working group for working through tutorials, bug fixes via github issues for Nvidia Nemo. Recruiting tech leads to work through self directed projects and demo as part of hands on learning process to build portfolio.
https://github.com/NVIDIA/NeMo
This is probably the fastest way to get involved in industry scale LLMs and agents. Review the NeMo tutorials for examples. We will start with using the A100 colab GPUs. Colab Pro is 9.99/month. This isn't paid for by a sponsor.
Sponsorship offer from Nvidia and AWS for free GPU time for projects leading to papers or source code contributions after qualification.
Requires ability to perform pull requests and debug. Does not require ML knowledge. Can learn everything else from debugging and doing demos.
We are going to try this for 2-3 months. There is a qualification process for this working group. Will message people who sign up and verify programming ability. I message you through meetup messages. Please make sure you can get those messages for first contact.
To stay in the group, the requirement is to do a demo in front of the group via zoom call. If working group members drop out then it isn't worth troubling corporate sponsors for GPU time. GPU sponsorship requires approval at VP level. We will try this for 2 months to see what the success rate is.
First meeting is intro and to see who can function as tech leads to lead meetups. Need volunteers in areas of computer vision, ASR, LLMs, distributed training, etc. Areas are listed in NeMo github.
Will require commitment to schedule and independently run meetups and do demos showing progression towards replicating tutorials and industry scale workflows.
Here are colab NeMo setup instructions.
https://github.com/dougc333/meetup_colab/blob/main/meetup_collab_cuda.ipynb
We are sold out. I can't handle more than 6 on a meetup call. We may start another session in a couple months after this ends.
