> Trained only on permissively-licensed repos to avoid legal risks
Do any of those “permissive” licences require attribution? Do they all grant a patent license?
Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
41–50 of 69 posts
Re: Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
#42Always astonishes me how negative hacker news can be whenever people try to launch stuff. This is cool!
When I read these threads I mentally prepend "this is neat" to all comments. Perhaps I'm too optimistic, but I think nearly any engagement with a new startup or idea is appreciation on some level.
Re: Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
#43Re: Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
#44How does this compare with https://www.tabnine.com/enterprise, which is also self-hosted, trained on permissively licensed repositories and supports training on private repos?
Re: Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
#45Always astonishes me how negative hacker news can be whenever people try to launch stuff. This is cool!
Negative reactions are quicker to arise and quicker to write out, which is a double whammy that causes them to appear more quickly and dominate the thread early. Slower, positive reactions eventually show up (usually!) and in the meantime we get the odd "contrarian dynamic" phenom where "I can't believe how negative these comments are!" objections start popping up and get upvoted to the top. Internet is weird.
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor...
Re: Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
#46Always astonishes me how negative hacker news can be whenever people try to launch stuff. This is cool!
Check out when Dropbox launched
Re: Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
#47Always astonishes me how negative hacker news can be whenever people try to launch stuff. This is cool!
Negative reactions are quicker to arise and quicker to write out, which is a double whammy that causes them to appear more quickly and dominate the thread early. Slower, positive reactions eventually show up (usually!) and in the meantime we get the odd "contrarian dynamic" phenom where "I can't believe how negative these comments are!" objections start popping up and get upvoted to the top. Internet is weird. https:…
The top comment usually goes like:
negative comments -> some presumably "funny" witty joke -> positive/meaningful insight -> peopel complain about comment quality
Re: Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
#48Always astonishes me how negative hacker news can be whenever people try to launch stuff. This is cool!
When I read these threads I mentally prepend "this is neat" to all comments. Perhaps I'm too optimistic, but I think nearly any engagement with a new startup or idea is appreciation on some level.
I usually prepend that to my critical comments ;)
If I’m responding in a critical manner it usually is because I think something is useful/ there.
Re: Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
#49Excited to try this looks really good! Been looking for something like this for our Pynecone code base!
Re: Launch HN: CodeComplete (YC W23) – Copilot for Enterprise
#50Have you benchmarked it against Copilot? Training data is only your own data, because that might not achieve a great accuracy? Does it work with any programming language?
We don't have any rigorous benchmarks against Copilot, but we're working on building an evaluation framework to do so. We've played a bunch with traditional academic metrics for codegen (e.g. pass @ k) and found they don't correlate super well with real-world performance. Also, want to mention that we are not competing directly with Copilot. A benchmark against Copilot is useful as we further improve our product, but…
What would a benchmark even look like?
Does it make useful code? Does it make the same code?
Or more strictly on something like latency and cost?