When your LLM agent is slow, is it the model? A tool call? The latest entry in my change seams series adds OpenTelemetry tracing and Prometheus metrics as a clean, swappable layer. #LLMOps#GenAI#Observability#AIEngineering
https://t.co/YKgQ8FYY9J
Model upgrades shouldn't be a two-day incident. I wrote about designing LLM apps with "change seams" - narrow interfaces that let you swap provider, prompt, tools, and observability independently.
https://t.co/xxhnMBq9vZ
@AzureSupport Is there a problem with the recognize_pii_entities service? It's failing (returning empty) if I run it on UK or France instances, but works in the US?
RedNevada's RAI is the best AI recruiting software in the market. It not only saves significant time on manual hiring processes, but also allows me to concentrate on other HR projects and contribute to the business's overall strategy https://t.co/DYtWQUuSV8
@OpenAI is there a problem with the moderation api? We've been seen a lot of 500 responses this afternoon.
Tested with the curl moderation example in the docs and also seeing the error.
For the past couple of weeks I've been playing with vector databases, the new way to store and query embedding data. https://t.co/kJIsGM5e0s #embedding#vectordatabase#weaviate#nlp
@BritishGasHelp I submitted my latest electricity reading via sms using the correct format. You have sent me nearly 100 messages in the past hour to say it has failed. Please stop sending me messages!
@ApacheSedona would it be possible to extend the functionality to n-dimensional spaces? I'm looking for a spark way to handle vector embeddings and quickly find similar/close points. https://t.co/UYjouJjgNG