I’m an instant pass on any startup that just wraps OpenAI. Zero differentiation over time.
Building your solution / app on your own models that you control, train and tune yourself? Good.
There's a new Open Source LLM model called "Cerebras-GPT" that hit the internet yesterday. I was curious how it compared in performance to #ChatGPT and #LLaMA so I wrote an article that compares them.
If you haven't heard of this yet, it's worth a read!
https://t.co/YYkVp8pOx2
Sneak peak of what we are teaching ChatGPT to do. Watch it browse the web like a human.
https://t.co/YFkW1gTBta
Amazing stuff. Not shown, we also have it hooked up to our vulnerability database. Should be available in a discord bot for you next week.
We could add this functionality into the dependency helper bot we've been building on GitHub. Please leave your thoughts in the comments below! (And if you'd like to try it out.)
A "truthy" lie will run 2x round the world faster than a boring truth.
The sad fact that a high effort piece from @alexrkonrad and @kenrickcai got overrun by lazy bullshit shows pulling the biggest possible number out of your ass while keeping a straight face is all you need.
So... @YCombinator has invested in 100+ COSS startups since 2008 (80%+ of those over the last 2 years alone!).
If this were abstracted out as a distinct fund, it would do extremely well... like a 100X fund, I think, and that's not an exaggeration. 😁
https://t.co/RVRu2VsTng
We've added environmental adjustment to LunaTrace. Answer a couple of questions about what type of app you're scanning and the severity of your vulnerabilities will be re-calibrated using the environmental CVSS spec, automatically.
https://t.co/xXIdVmuBH3
Status update for 2023: Here is a video of @breadchris showing off the new dynamic tracing capabilities that we've been adding to LunaTrace over the past few weeks.
You can now see which CVEs are ever executed in production-- a helpful signal for patching!
https://t.co/nw0d0iT2hF