cjamsonhn

5 ms·

1. 1.▲
   by cjamsonhn 10d ago

   It would interesting to leave a trail of prompts to try to get them do things for you
2. 2.▲
   by cjamsonhn 10d ago

   you're absolutely right!
3. 3.▲
   Intuitions for Tranformer Circuits(connorjdavis.com)

   67 points by cjamsonhn 6mo ago|5 comments
4. 4.▲
   by cjamsonhn 6mo ago

   >I was smart enough to realize I'd never be a good researcher relatively early during my PhD and started preparing for job interviews What are some things that made you realize that? Im considering going into research and certainly
5. 5.▲
   Language Modeling, Part 5: Reverse Engineering LSTM Cells(connorjdavis.substack.com)

   1 points by cjamsonhn 8mo ago|0 comments
6. 6.▲
   Language Modeling Part 4: LSTMs(connorjdavis.substack.com)

   3 points by cjamsonhn 8mo ago|0 comments
7. 7.▲
   Langauge Modeling, Part 3: Vanilla RNNs(connorjdavis.substack.com)

   1 points by cjamsonhn 8mo ago|0 comments
8. 8.▲
   Language Modeling, Part 2: Training Dynamics(connorjdavis.substack.com)

   1 points by cjamsonhn 8mo ago|0 comments
9. 9.▲
   by cjamsonhn 9mo ago

   Highly recommend this as well. Does a great job of helping you build intuition for why things like gradient descent and normalization work. Also gets into the weeds on training dynamics and how to ensure they are behaving properly
10. 10.▲
    by cjamsonhn 9mo ago

    > Because the motherfuckers behind public AI interfaces fine-tune them to be as human-like, as rewarding, as dopamine-inducing, as addictive, as possible? And to think they dont even have ad-driven business models yet