21–30 of 134 posts
GPT-4 cost like $100m so I don't think this is surprising?
A lot of organizations still think they should have their own [finetuned] model to provide a custom experience to their users, so that may come as a surprise for them.
Extremely hot LLM take: You will often get better results with few-shot prompting (with good examples) on a modern LLM than with a finetuned LLM. Finetuning was the best option for weaker LLMs with lower context windows (e.g. the original GPT-3): both problems have been solved nowadays. The cost economics are much better with few-shot prompting to modern LLMs too: input tokens are super cheap (especially with the rec…
Fine tuning will disappear, no reason to invest so heavily in it. Also, most LLM code being written right now (prompts, RAG, etc) will be obsolete once the next model comes out. Anyone starting an LLM application startup is arguably wasting their time, wait until the next iteration is out. Then you will know whats possible.
Not true. Most prompt techniques that work on current modern LLM models will work on different or future models, although it will require a QA pass for any regressions.
Earlier quoted context omitted.
A lot of organizations still think they should have their own [finetuned] model to provide a custom experience to their users, so that may come as a surprise for them.
Scaling laws basically guarantee that a sufficiently larger general model will usually beat a smaller specialist model. The misunderstanding is perhaps acceptable but the headline here is essentially restating a well known property of deep learning.
Does anything currently beat GPT-4? I saw some comments here say to check out Claude. From what I can tell, Claude hasn't figure out yet how to do the whole "generate Python code and run it in a Juptyer notebook" for math yet.
Fine tuning will disappear, no reason to invest so heavily in it. Also, most LLM code being written right now (prompts, RAG, etc) will be obsolete once the next model comes out. Anyone starting an LLM application startup is arguably wasting their time, wait until the next iteration is out. Then you will know whats possible.
> Also, most LLM code being written right now (prompts, RAG, etc) will be obsolete once the next model comes out. Not true. Most prompt techniques that work on current modern LLM models will work on different or future models, although it will require a QA pass for any regressions.
Fine tuning will disappear, no reason to invest so heavily in it. Also, most LLM code being written right now (prompts, RAG, etc) will be obsolete once the next model comes out. Anyone starting an LLM application startup is arguably wasting their time, wait until the next iteration is out. Then you will know whats possible.
Extremely hot LLM take: You will often get better results with few-shot prompting (with good examples) on a modern LLM than with a finetuned LLM. Finetuning was the best option for weaker LLMs with lower context windows (e.g. the original GPT-3): both problems have been solved nowadays. The cost economics are much better with few-shot prompting to modern LLMs too: input tokens are super cheap (especially with the rec…
can you give a few pointers on articles or examples of this?
You are an expert copywriter. Write five distinct blog post titles optimized for high clickthrough for Hacker News for the article the user provides.
Your response must follow the style of these titles:
- The ü/ü Conundrum
- Why isn't preprint review being adopted?
- Majority of web apps could just run on a single server
- Weather Planning for Eclipse Day
- PSChess – A chess engine in PostScript
I just ran one of my blog posts (https://news.ycombinator.com/item?id=39495476) with the workflow through Claude Haiku and got this:
Here are five distinct blog post titles optimized for high clickthrough on Hacker News for the article provided:
1. Tipping ChatGPT: Does Offering Monetary Incentives Improve AI Text Generation?
2. Quantifying the Impact of Incentives on Large Language Model Performance
3. Carrot or Stick? Exploring the Effects of Positive and Negative Prompts on ChatGPT
4. Gamifying AI: Using "Generation Golf" to Test ChatGPT's Ability to Follow Length Constraints
5. The Curious Case of ChatGPT's Motivations: Can an AI Be Incentivized Like Humans?