For instance used it in conjunction with llama-index for knowledge management. Created an index for a whole confluence/jira of a mid-sized company, got good results with GPT, but for LLaMA of this size that use case was too much.
I tested on a serious use case and quality was subpar. For real use cases I had to either host the most powerful model you can get (e.g. LLaMA-65B or so) on a cloud machine, which again costs too much (you'll be paying like 500-1000 USD per month), or just go straight for GPT-3.5 on OpenAI. The latter economically makes most sense.
yea, you're right. It's really pre-MVP. Basically an API in between your user facing input and OpenAI that detects prompt injections and flags them for you so you can abort sending to OpenAI.
I know an amazing IC at a software company I once worked at that I still keep in touch with. He has extremely deep technical knowledge, which you can simply deduct from basically everyone in the company (even from other teams) coming to him for advice. He's been at the company for >10 years.
He has strong opinions on current processes and just getting things done.
This post resonated deeply with me, since I've discussed before with him his role and how it could evolve. I know for a fact that he dislikes lots of meeting and really likes working on the core product, and so far hasn't really jumped on the opportunity to go into management - he doesn't really want to be manager. So he is kind of exactly the guy the post is describing. The company is growing though and he is very slowly getting pushed by the head of development into a more managerial role..
Let's see how it works out. I believe he is going to be a great manager though.
To add to this, as a German, the future of Uber does not look great in Europe as well. The taxi lobby is strong here, and I’m sure regulations will come in countries like Germany and will affect negatively their business. And they're starting to face stronger competition from other companies like Bolt.
I've been collecting these for the past year - there is a large body of research going into solving optimization problems using quantum computing, and some interesting POC results coming out using the latest machines.
I wouldn't say that's a YC-specific issue. Look at Nasdaq, this is a general issue with tech stocks right now, just this immense selling all across the board.
Putting things in relative like that (e.g. 30% more likely to be murdered) always feels like much - in reality, one doesn't know the variance. This is just putting the averages of two datasets in relation. If both have a very wide confidence interval, 30% suddenly loses meaning to me..
Hello HN! This is a SaaS I've built the last few months using Flask and the least amount of JS that is possible.
It allows setting up SMS/E-Mail/App alerts for stocks, options and crypto, and in the future also bonds and macroeconomic events. There are lots of KPIs (just for example: earnings growth Q/Q, price, volume, the option greeks, etc).
For stocks the main data source right now is Yahoo Finance, but at some point I'd like to move to a better one (but more costly one).
Since there are quality differences between data providers, it would be good to know where you source the data from. Is it from Yahoo Finance (just for example)? Or is it really directly from Nasdaq (which I presume would be pretty costly, and thus raises the question on how long you'll be able to run this service for free before shutting down or having to raise prices).
This is important since if someone decides to fully build on top of your API, they don't want the service to be gone in a few months and waste all the time integrating with yours.
I've checked the original source [0]: You've guessed right. They specifically said they won't be introducing/creating new gas models. Nowhere do they say that they won't sell old inventory.
I was annoyed by the seemingly over-engineered tools for screen tracking, so I wrote one myself in ~50 LOC, which simply uses ffmpeg to create a screenshot every X seconds in a very low resolution: