I feel like the explicitly go out of their away to avoid building in custom instructions/increased prompt context to lower their costs so this is a bit of a work around.
copilot scans open files so I mention the file name in my prompt or reference the file explicitly using the reference file option in the chat window in the intellij copilot chat beta.
```
Follow the instructions for preparing responses from the COPILOT.md file in how you answer and format your answer to the following prompt.
I came across a nice little solution in copilot X and copilot 365/pro for including and referencing custom instructions to add behavior/formatting support to copilot (e.g. edit mode, article writing mode, documentation writting mode etc.)
Which is fundamentally different from how our brain chains together thoughts when not actively engaging in meta thinking how? Especially once chain of thought etc. is applied.
You can tell GPT to output sentiment analysis and mind reading of user intent, what it believes the user's underlying goal is. It becomes less responsive if it finds the user to not be friendly or perceives them as trying to achieve a restricted outcome.
I was really blown away by how well erlang/chicago boss handled throughput when I first gave it a spin. A webpage I converted over handled multiple orders of magnitude more incoming requests before failing than the php version it replaced.
CB would get frustrating when wanting to extend certain functionality however, which phoenix isn't quite as bad about.
yep, prepriming the cache rather than passively allowing it be rebuilt by request/queries can also result in some nice improvements and depending on replication delay across database servers avoid some unexpected query results reaching the end user.
In the past I was the architect of a top 2000 alexa ranked social networking site, data synchronization delays were insane under certain load patterns high single low double digit second write propagation delays.
Yes.
I handle around a million requests per minute. I exponentially increase the cache period after subsequent misses to avoid an outage ddos the whole system.
This tends to be beneficial regardless of the root cause.
edit this is especially useful for handling search/query misses as a query with no results is going to scan any relevant indexes etc. until it is clear no match exists meaning a no results query may take up more cycles than a hit.
print some spoof news articles periodically and sue them when they duplicate your "parody content" like the yellow books used to do to prevent duplication of phone numbers.
True, but even if we were say 80% confident in each rebuttal being the true reason chained together the odds of those all being correct is pretty small. Although of course it could be due to alternative rebuttals being the correct ones.
At a certain point in time, not that we've reached it, it starts to look like the ptolemaic system, or luminiferous aether, where you keep tacking on exceptions or explanations to fit the data to your model instead of your model to the data.