@unclebobmartin It also started sending touch commands via CLI to switch screens etc.
Sometimes it hallucinates visual results on the screenshots though, so there's obviously room for improvement.
@unclebobmartin I'm doing a hobby project on android where I have Codex screenshot the UI after each iteration and then have it decide whether it wants to continue making changes or get back to me for feedback. Already had it do 4-5 iterations until it was happy with the result and check back.
@ID_AA_Carmack Agree. Single assignment lets me mark a value as stable across a function/class and move on. It reduces the mental bookkeeping; mutation should stand out.
I've been asked a few times recently about setting up container-to-container communication using #testcontainers for #dotnet. One example, aside from our docs, is our @FlywayDb example where the Flyway container runs the DB migration before running tests: https://t.co/K3qttzzQTP
I haven't shared much about recent #testcontainers for #dotnet releases, but I’d like to highlight some key features in this release: automatic container runtime discovery from context and image pinning by digest. Thanks to the contributors 🙏 https://t.co/enOnVCww4O
High quality integration tests are worth 10x any unit test with mocks.
We've been slowly adding more of these in conjunction with testcontainers in favor of mocked unit tests (where relevant).
The tests can take a little longer, but the confidence they provide is well worth it.
Ollama 0.3 with tool support!
You can now use tool calling with popular models such as Llama 3.1!
👇👇👇
https://t.co/T3yFy5HRO6
Example tools include:
- Functions & APIs
- Web browsing
- Code interpreter
- and much more!
🧵 quick thread
Did the calculator make the cost of math go to zero? I don't even know what that means, but just now that we have a tool that crunches words and not numbers, I'm not sure that it will drastically change the value of things.