long-range-arena
attention-is-all-you-need-pytorch
Our great sponsors
long-range-arena | attention-is-all-you-need-pytorch | |
---|---|---|
6 | 3 | |
682 | 8,456 | |
2.9% | - | |
0.0 | 0.0 | |
4 months ago | 11 days ago | |
Python | Python | |
Apache License 2.0 | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
long-range-arena
-
The Secret Sauce behind 100K context window in LLMs: all tricks in one place
https://github.com/google-research/long-range-arena
-
[R] The Annotated S4: Efficiently Modeling Long Sequences with Structured State Spaces
The Structured State Space for Sequence Modeling (S4) architecture is a new approach to very long-range sequence modeling tasks for vision, language, and audio, showing a capacity to capture dependencies over tens of thousands of steps. Especially impressive are the model’s results on the challenging Long Range Arena benchmark, showing an ability to reason over sequences of up to 16,000+ elements with high accuracy.
-
[D] Is there a repo on which many light-weight self-attention mechanism are introduced?
1.1 Long Range Arena: A Benchmark for Efficient Transformers. From authors of above, they proposed a benchmark for modeling long range interactions. It also inlcudes a repository
- [R] Google’s H-Transformer-1D: Fast One-Dimensional Hierarchical Attention With Linear Complexity for Long Sequence Processing
- [2107.11906] H-Transformer-1D: Fast One-Dimensional Hierarchical Attention for Sequences
-
[R][D] Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting. Zhou et al. AAAI21 Best Paper. ProbSparse self-attention reduces complexity to O(nlogn), generative style decoder to obtainsequence output in one step, and self-attention distilling for further reducing memory
I think the paper is written in a clear style and I like that the authors included many experiments, including hyperparameter effects, ablations and extensive baseline comparisons. One thing I would have liked is them comparing their Informer to more efficient transformers (they compared only against logtrans and reformer) using the LRA (https://github.com/google-research/long-range-arena) benchmark.
attention-is-all-you-need-pytorch
-
ElevenLabs Launches Voice Translation Tool to Break Down Language Barriers
The transformer model was invented to attend to context over the entire sequence length. Look at how the original authors used the Transformer for NMT in the original Vaswani et al publication. https://github.com/jadore801120/attention-is-all-you-need-py...
-
Question: LLMs
I did implement an "LLM" proof of concept from scratch in a course for my masters, pretty much doing a small implementation of a transformer from the Attention is all you Need paper (plus other resources). It was useless, but was a great experience to understand how it works. There are a few implementation like this out there, like this one: https://github.com/jadore801120/attention-is-all-you-need-pytorch (first google result). I think it is a fun exercise (the amount of fun depends on how much of a masochist you are :) ).
-
Lack of activation in transformer feedforward layer?
I'm curious as to why the second matrix multiplication is not followed by an activation unlike the first one. Is there any particular reason why a non-linearity would be trivial or even avoided in the second operation? For reference, variations of this can be witnessed in a number of different implementations, including BERT-pytorch and attention-is-all-you-need-pytorch.
What are some alternatives?
performer-pytorch - An implementation of Performer, a linear attention-based transformer, in Pytorch
LFattNet - Attention-based View Selection Networks for Light-field Disparity Estimation
HJxB - Continuous-Time/State/Action Fitted Value Iteration via Hamilton-Jacobi-Bellman (HJB)
BERT-pytorch - Google AI 2018 BERT pytorch implementation
jax-resnet - Implementations and checkpoints for ResNet, Wide ResNet, ResNeXt, ResNet-D, and ResNeSt in JAX (Flax).
OpenPrompt - An Open-Source Framework for Prompt-Learning.
tldr-transformers - The "tl;dr" on a few notable transformer papers (pre-2022).
allennlp - An open-source NLP research library, built on PyTorch.
elegy - A High Level API for Deep Learning in JAX
transformer-pytorch - Transformer: PyTorch Implementation of "Attention Is All You Need"
transformers - 🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.