News for hackers. Stuff that matters.
Re:A Philosophical Introduction to Language Models by visarga on Thursday January 11, 2024 @04:03PM (#38959045)
This is the "Anti Stochastic Parrot" paper. I am happy to finally read a philosophical paper on LLMs.

I just have a small quip.

> While LLMs show promise in various forms of task generalization, their participation in the ratcheting process of cultural learning thus appears contingent on further advancements in these areas, which might lie beyond the reach of current architectures.

I don't think cultural transmission with LLMs is that incipient.

Evolution through Large Models - https://arxiv.org/abs/2206.08896

We have seen papers where LLMs are used as "search heads" or "mutation operators" in evolutionary methods. They can rely on their language proficiency to explore the most promising leads, spanning a vast combinatorial space.

And another example of "cultural transmission": GPT-4 has its paws all over Mistral and LLaMA finetunes. Plenty of cultural transmission between big and small LLMs. Not to mention Phi, a model trained purely on synthetic GPT-4 data. Synthetic data has been estimated to be 5x more efficient than human data, you can make a model 1/5 size with similar performance.

Re:A Philosophical Introduction to Language Models by ShamelessC on Thursday January 11, 2024 @04:28PM (#38959351)
> Synthetic data has been estimated to be 5x more efficient than human data, you can make a model 1/5 size with similar performance.

Interesting. Can you cite anything to back up that claim?

Re:A Philosophical Introduction to Language Models by visarga on Thursday January 11, 2024 @04:31PM (#38959401)
Yes.

> We follow the “Textbooks Are All You Need” approach, focusing this time on common sense reasoning in natural language, and create a new 1.3 billion parameter model named phi-1.5, with performance on natural language tasks comparable to models 5x larger, and surpassing most non-frontier LLMs on more complex reasoning tasks such as grade-school mathematics and basic coding.

https://arxiv.org/pdf/2309.05463.pdf

Re:A Philosophical Introduction to Language Models by ShamelessC on Thursday January 11, 2024 @05:46PM (#38960534)
Thanks!
Re:A Philosophical Introduction to Language Models by ignoramous on Thursday January 11, 2024 @04:58PM (#38959843)
See also Self-Play Fine-Tuning (SPIN) https://arxiv.org/abs/2401.01335 that (per authors) tunes as well as Direct Preference Optimization (DPO).