composer VS Pytorch

Mosaic's open source library is excellent: Composer https://github.com/mosaicml/composer.

* It gives you PyTorch DDP for free. Makes FSDP about as easy as can be, and provides best in class performance monitoring tools. https://docs.mosaicml.com/en/v0.12.1/notes/distributed_train...

Here's a nice intro to using Huggingface models: https://docs.mosaicml.com/en/v0.12.1/examples/finetune_huggi...

I'm just a huge fan of their developer experience. It's up there with Transformers and Datasets as the nicest tools to use.

You may consider using Composer Composer by MosaicML.

Which is why I'm excited to announce that we (MosaicML) just released an automatic way to avoid these errors. Namely, we just added automatic gradient accumulation to Composer, our open source library for faster + easier neural net training.

I would love to know your thoughts on PyTorch Lightning vs. other, even more lightweight libraries, if you have the time. PL strikes me as being less idiosyncratic than FastAI, but I'm still not sure whether it would be better in engineering work to go even more lightweight (when I'm not just writing the code myself) -- something that offers up just optimizations and a trainer, a la MosaicML's [Composer](https://github.com/mosaicml/composer) or Chris Hughes's [pytorch-accelerated](https://github.com/Chris-hughes10/pytorch-accelerated) .

This master's thesis sort of does it, but it doesn't have any fine-tuning yet so it completely wrecks the accuracy: https://github.com/joennlae/halutmatmul.

If someone worked on contributing this to Composer [1] I'd be down to help out. I can't justify building it all on my own right now since we're 100% focused on training speedup, but I could definitely meet and talk through it, help code tricky parts, review PRs, etc.

[1] https://github.com/mosaicml/composer

We're always looking for contributors for Composer. tl;dr it speeds up neural net training by a lot (e.g., 7x faster ResNet-50).

Looking at this: https://github.com/mosaicml/composer

torch is a package to manage tensors and dynamic neural networks in python (GitHub)

PyTorch also has some support for them, but it's quite incomplete and has many issues so that it is basically unusable. And its future development is also unclear. https://github.com/pytorch/pytorch/issues/60832

TensorFlow

TensorFlow, developed by Google, and PyTorch, developed by Facebook, are two of the most popular frameworks for building and training complex machine learning models. TensorFlow is known for its flexibility and robust scalability, making it suitable for both research prototypes and production deployments. PyTorch is praised for its ease of use, simplicity, and dynamic computational graph that allows for more intuitive coding of complex AI models. Both frameworks support a wide range of AI models, from simple linear regression to complex deep neural networks.

> does PyTorch have a similar concept

of course https://github.com/pytorch/pytorch/blob/main/torch/utils/_py...

fyi should work on most 40xx[1]

[1] https://github.com/pytorch/pytorch/issues/119638#issuecommen...

Sure, right here: https://github.com/pytorch/pytorch/blob/main/torch/autograd/...

Here's the documentation: https://pytorch.org/tutorials/intermediate/forward_ad_usage....

> When an input, which we call “primal”, is associated with a “direction” tensor, which we call “tangent”, the resultant new tensor object is called a “dual tensor” for its connection to dual numbers[0].

*My post explains Dot, Matrix and Element-wise multiplication in PyTorch.

In PyTorch with @, dot() or matmul():

import torch # we use PyTorch: https://pytorch.org data = torch.tensor(encode(text), dtype=torch.long) print(data.shape, data.dtype) print(data[:1000]) # the 1000 characters we looked at earlier will to the GPT look like this