bertviz VS transformers

This seems only to work cause large GPTs have redundant, undercomplex attentions. See this issue in BertViz about attention in Llama: https://github.com/jessevig/bertviz/issues/128

https://github.com/jessevig/bertviz this could be helpful .. I was playing around with it a while ago to see how the attention weights are distributed across prompts

It would be interesting to have attention visualized as well, similar to how it's done in BertViz:

https://github.com/jessevig/bertviz

link for lazy: https://github.com/jessevig/bertviz

The Illustrated Transfomer ( https://jalammar.github.io/illustrated-transformer/ ) and Visualizing attention ( https://towardsdatascience.com/deconstructing-bert-part-2-vi... ), are both really good resources. For a more ELI5 approach this non-technical explainer ( https://www.parand.com/a-non-technical-explanation-of-chatgp... ) covers it at a high level.

You can get some information this way, but not everything you would want to know. You can try it yourself with BertViz.

2) BERT learns a lot in its embeddings: the BERTOLOGY paper (https://arxiv.org/abs/2002.12327) provides a great in-depth look at some of the broader linguistic traits that BERT learns. Different layers often learn different patterns, so the embeddings aren't really interpretable, but you can use something like bertviz (https://github.com/jessevig/bertviz) to explore attention weights across layers for predetermined examples

I messed up a few things and mixed a couple others, anyways this site has a lot of decent information about it. https://towardsdatascience.com/deconstructing-bert-part-2-visualizing-the-inner-workings-of-attention-60a16d86b5c1

Big fan of BertViz for this, widely used in research for this very purpose: https://github.com/jessevig/bertviz

transformers is a package by Hugging Face, that helps you interact with models on HF Hub (GitHub)

Is t5x an encoder/decoder architecture?

Some more general options.

The Flax ecosystem

https://github.com/google/flax?tab=readme-ov-file

or dm-haiku

https://github.com/google-deepmind/dm-haiku

were some of the best developed communities in the Jax AI field

Perhaps the “trax” repo? https://github.com/google/trax

Some HF examples https://github.com/huggingface/transformers/tree/main/exampl...

Sadly it seems much of the work is proprietary these days, but one example could be Grok-1, if you customize the details. https://github.com/xai-org/grok-1/blob/main/run.py

The HuggingFace transformers library already has support for a similar method called prompt lookup decoding that uses the existing context to generate an ngram model: https://github.com/huggingface/transformers/issues/27722

I don't think it would be that hard to switch it out for a pretrained ngram model.

Most of this tutorial is based on Hugging Face course about Transformers and on Niels Rogge's Transformers tutorials: make sure to check their work and give them a star on GitHub, if you please ❤️

* Superconvergence + LR range finder + Fast AI's Ranger21 optimizer was the goto optimizer for CNNs, and worked fabulously well, but on transformers, the learning rate range finder sadi 1e-3 was the best, whilst 1e-5 was better. However, the 1 cycle learning rate stuck. https://github.com/huggingface/transformers/issues/16013

Thanks! :) I'm pushing them into transformers, pytorch-gemma and collabing with the Gemma team to resolve all the issues :)

The RoPE fix should already be in transformers 4.38.2: https://github.com/huggingface/transformers/pull/29285

My main PR for transformers which fixes most of the issues (some still left): https://github.com/huggingface/transformers/pull/29402

If you want to tinker with the architecture Hugging Face has a FOSS implementation in transformers: https://github.com/huggingface/transformers/blob/main/src/tr...

If you want to reproduce the training pipeline, you couldn't do that even if you wanted to because you don't have access to thousands of A100s.