RNNs vs Transformers

The cluster centers on debates about whether a new neural architecture reinvents RNNs or LSTMs, with comparisons to transformers, discussions on their strengths, weaknesses, and modern alternatives like Mamba.

📉 Falling 0.5x AI & Machine Learning
1,909
Comments
19
Years Active
5
Top Authors
#4911
Topic ID

Activity Over Time

2007
1
2009
3
2010
3
2011
4
2012
2
2013
8
2014
25
2015
148
2016
229
2017
262
2018
183
2019
171
2020
86
2021
61
2022
57
2023
244
2024
282
2025
136
2026
4

Keywords

e.g RNN LLM TensorFlow LSTM FAIR youtube.com CNN IMO FastText transformer transformers neural architecture tensorflow input neural network models nets architectures

Sample Comments

wantsanagent • Oct 11, 2024 • View on HN

Someone explain to me how this isn't reinventing LSTMs please.

dnautics • Nov 10, 2021 • View on HN

yeesh it's an RNN and not even something like a transformer

iAkashPaul • Nov 13, 2024 • View on HN

Could be that or an RNN/LSTM as well

dekhn • Jul 1, 2024 • View on HN

Are RNNs completely subsumed by transformers? IE, can I forget about learning anything about how to work with RNNs, and instead focus on transformers?

quelltext • Jul 23, 2019 • View on HN

Please elaborate. What's wrong with RNNs and what should be used instead?

halflings • Jan 5, 2017 • View on HN

You're right, that does look similar... I expected this to be based on some type of RNN!

dbagr • Nov 28, 2024 • View on HN

This sounds like an RNN with extra steps.

caf • Jul 13, 2016 • View on HN

This seems like perfect fodder for an RNN.

jdeaton • Dec 20, 2023 • View on HN

Very cool ive read this line of paper originating from hippo, s4, hyena, mamba etc but can someone please explain how this isnt just an RNN/LSTM variant??

richardsocher • Dec 21, 2021 • View on HN

Unless he models are recursive neural networks :)