metaseq VS minGPT

This is a super important issue that affects the pace and breadth of iteration of AI almost as much as the raw hardware improvements do. The blog is fun but somewhat shallow and not technical or very surprising if you’ve worked with clusters of GPUs in any capacity over the years. (I liked the perspective of a former googler, but I’m not sure why past colleagues would recommend Jax over pytorch for LLMs outside of Google.) I hope this newco eventually releases a more technical report about their training adventures, like the PDF file here: https://github.com/facebookresearch/metaseq/tree/main/projec...

The number of people who can actually pre-train a true LLM is very small.

It remains a major feat with many tweaks and tricks. Case in point: the 114 pages of OPT175B logbook [1]

[1] https://github.com/facebookresearch/metaseq/blob/main/projec...

Reliability at scale: take a look at the OPT training log book for their 175B model run. It needed a lot of babysitting. In my experience, that scale of TPU training run requires a restart about once every 1-2 weeks—and they provide the middleware to monitor the health of the cluster and pick up on hardware failures.

I really appreciated this log of Facebook training a large language model of how troublesome AI development can be: https://github.com/facebookresearch/metaseq/tree/main/projects/OPT/chronicles

Stable Diffusion will run on any decent gaming GPU or a modern MacBook, meanwhile LLMs comparable to GPT-3/ChatGPT have had pretty insane memory requirements - e.g., <https://github.com/facebookresearch/metaseq/issues/146>

It seems so, check this log book from Meta: https://github.com/facebookresearch/metaseq/blob/main/projec...

minGPT (Karpathy): https://github.com/karpathy/minGPT

Next, some foundational textbooks for general ML and deep learning:

Language Model novice here. I was going through the README section of minGPT and read this line.

The first network displayed with working weights is a tiny such network, which sorts a small list of the letters A, B, and C. This is the demo example model from Andrej Karpathy's minGPT implementation.

Tried it once. Its promise is to take the dataset's seasonal trend into account, which makes sense for Facebook's original use case.

We ran it on such a dataset and found out that directly using https://github.com/karpathy/minGPT consistently gives a better result. So we ended up using the output of Prophet as an input feature to a neural network, but the result was not improved in any significant way.

Sure, im trying to regenerate a bunch of complex numbers based on their absolute value. So im trying to embed these absolute values and then using gpt model(probably mini gpt) try to recover the original comples numbers. There is a certain connection between these complex numbers and their order which im not capable of explaining yet. Im hoping the model would be capable of recognizing certain sequences of these absolute values and match them with the desired complex counterparts (by training the model).

minGPT (https://github.com/karpathy/minGPT)

Check out minGPT and nanoGPT from Karpathy, he puts out some of the best machine learning tutorials and teaching content.