@eify.bsky.social
I just noticed this: Did Meta AI Few-Shot Learner's use of policy description inspire Anthropic's Constitutional AI, a year later?
ai.meta.com/blog/harmful...
www.anthropic.com/research/con...
ai.meta.com/blog/harmful...
www.anthropic.com/research/con...
Harmful content can evolve quickly. Our new AI system adapts to tackle it.
We’ve built and deployed a new AI technology called Few-Shot Learner that can take faster action on new or evolving types of harmful content.
ai.meta.com
January 1, 2025 at 12:56 AM
I just noticed this: Did Meta AI Few-Shot Learner's use of policy description inspire Anthropic's Constitutional AI, a year later?
ai.meta.com/blog/harmful...
www.anthropic.com/research/con...
ai.meta.com/blog/harmful...
www.anthropic.com/research/con...
It seems that AdamW & MARS (arxiv.org/abs/2411.10438) effectively reach the same val loss for GPT-2 small with the optimal LR according to Appendix B, in contrast to Figure 1? @quanquangu.bsky.social
If MARS is less sensitive to LR that's also an advantage, but a different kind.
If MARS is less sensitive to LR that's also an advantage, but a different kind.
December 4, 2024 at 9:31 PM
It seems that AdamW & MARS (arxiv.org/abs/2411.10438) effectively reach the same val loss for GPT-2 small with the optimal LR according to Appendix B, in contrast to Figure 1? @quanquangu.bsky.social
If MARS is less sensitive to LR that's also an advantage, but a different kind.
If MARS is less sensitive to LR that's also an advantage, but a different kind.
Critical module of TensorFlow (TensorFlow Text) still doesn't support Python 3.12. I had to switch back to 3.11 😬
github.com/tensorflow/t...
github.com/tensorflow/t...
GitHub - tensorflow/text: Making text a first-class citizen in TensorFlow.
Making text a first-class citizen in TensorFlow. Contribute to tensorflow/text development by creating an account on GitHub.
github.com
November 28, 2024 at 7:44 AM
Critical module of TensorFlow (TensorFlow Text) still doesn't support Python 3.12. I had to switch back to 3.11 😬
github.com/tensorflow/t...
github.com/tensorflow/t...
That training run failed to converge anyway but TIL if you use multiprocessing to spawn new processes you can't edit .py file while the code is running: python interpreter needs to compile the source again to do that!
November 25, 2024 at 8:41 PM
That training run failed to converge anyway but TIL if you use multiprocessing to spawn new processes you can't edit .py file while the code is running: python interpreter needs to compile the source again to do that!