Anonymous info

Tuesday, September 12, 2023

New best story on Hacker News: Ask HN: Why did Visual Basic die?

September 12, 2023 Hacker News No comments

Ask HN: Why did Visual Basic die?
405 by Zelphyr | 507 comments on Hacker News.
I've been a software developer for almost 30 years. I remember using VB back in the 90's and I was thinking about it the other day and it dawned on me; despite all the advances in technology since then, nothing I have found compares to that development experience today. I would go so far as to say we've gone backwards in a big way. Now, I'm no fan of Microsoft products but, I have yet to find a tool that can allow me to be as productive in so short a time as Visual Basic. Yet I can't help wondering what problems it had that caused them to abandon it? Moreover, why hasn't someone come out with a solid replacement?

New best story on Hacker News: Fine-tune your own Llama 2 to replace GPT-3.5/4

September 12, 2023 Hacker News No comments

Fine-tune your own Llama 2 to replace GPT-3.5/4
487 by kcorbitt | 125 comments on Hacker News.
There has been a lot of interest on HN in fine-tuning open-source LLMs recently (eg. Anyscale's post at https://ift.tt/NK6SEwm ). I've been playing around with fine-tuning models for a couple of years, and wanted to share some insights and practical code. I’ve condensed what I’ve learned into a small set of notebooks at https://ift.tt/FYtNnTf... , covering labeling data, fine-tuning, running efficient inference, and evaluating costs/performance. The 7B model we train here matches GPT-4’s labels 95% of the time on the test set, and for the 5% of cases where they disagree it’s often because the correct answer is genuinely ambiguous. What is fine-tuning? You can think of it as a more-powerful form of prompting, where instead of writing your instructions in text you actually encode them in the weights of the model itself. You do this by training an existing model on example input/output pairs that demonstrate the task you want your fine-tuned model to learn. Fine-tuning can work with as few as 50 examples but I usually try to get 1000+ if possible. Prompting still has some big advantages over fine-tuning. It's way easier/faster to iterate on your instructions than label data and re-train a model. And operationally it's easier to deploy one big model and just adjust its behavior as necessary vs deploying many small fine-tuned models that will likely each get lower utilization. Fine-tuning has one huge advantage though: it is far more effective at guiding a model's behavior than prompting, so you can often get away with a much smaller model. That gets you faster responses and lower inference costs. A fine-tuned Llama 7B model is 50x cheaper than GPT-3.5 on a per-token basis, and for many use cases can produce results that are as good or better! For example, classifying the 2M recipes at https://ift.tt/T3la7Kp with GPT-4 would cost $23k. Even with GPT-3.5 it would cost over $1k. The model we fine-tuned performs similarly to GPT-4 and costs just $19 to run over the entire dataset. Disclaimer: My brother David and I are working on an open-source product called OpenPipe ( https://ift.tt/Rt3oEqZ ) to help engineers adopt fine-tuning as simply as possible. But none of the information above depends on our startup. The current post is just about sharing information that we’ve learned about fine-tuning. I hope it’s useful!

Anonymous info

This is default featured slide 1 title

This is default featured slide 2 title

This is default featured slide 3 title

This is default featured slide 4 title

This is default featured slide 5 title

Tuesday, September 12, 2023

New best story on Hacker News: Ask HN: Why did Visual Basic die?

New best story on Hacker News: Fine-tune your own Llama 2 to replace GPT-3.5/4

New best story on Hacker News: Death by a Thousand Microservices

Monday, September 11, 2023

New best story on Hacker News: Every type of plastic used by LEGO (2022)

New best story on Hacker News: Google Chrome just rolled out a new way to track you and serve ads

New best story on Hacker News: Aardvark'd: The Fog Creek documentary, 18 years later

New best story on Hacker News: Nginx Unit – Universal web app server

Sunday, September 10, 2023

New best story on Hacker News: Lithium discovery in US volcano could be biggest deposit ever found

New best story on Hacker News: Asking 60 LLMs a set of 20 questions

New best story on Hacker News: Privacy Nightmare on Wheels’: Every Car Brand Reviewed by Mozilla

Friday, September 8, 2023

New best story on Hacker News: Maybe Rust isn’t a good tool for massively concurrent, userspace software

New best story on Hacker News: Touch Pianist

New best story on Hacker News: Bun v1.0.0

Thursday, September 7, 2023

New best story on Hacker News: Mullvad on Tailscale: Privately browse the web

New best story on Hacker News: NSO group iPhone zero-click, zero-day exploit captured in the wild

New best story on Hacker News: Kagi Small Web

New best story on Hacker News: Show HN: Host a Website in the URL

Wednesday, September 6, 2023

New best story on Hacker News: U.K. abandons, for now, legislation that would have banned end-to-end encryption

New best story on Hacker News: Internet-connected cars fail privacy and security tests conducted by Mozilla

New best story on Hacker News: Is this Duplo train track under too much tension?

New best story on Hacker News: Learn WebAssembly by writing small programs

Labels

Blog Archive

Blogger templates

Labels

Blogroll

About