@duolingo this new update sucks. Not cool that even if you haven't made a mistake you're forced to pay to continue. Education should not be this cutthroat.
Quite an interesting publication by @huggingface team. Reflecting on it from an explanability perspective, I worry AI developments today are heading in the wrong direction. There is less visibility on causal links so who do we blame? Dirty data? https://t.co/O3yaXf58lF
@JCrawford1975@mjovanovictech He means no harm. I think the goal is that you can make calls to the database to quickly validate the request from fluent validation.
@ljquintanilla@mjovanovictech Really awesome video from @mjovanovictech, I must say. Unfortunately, it seems IChatClient was updated. I also explored it using Deepseek r1 if you want to check it out here: https://t.co/SGDskcguMt
Spent a couple of hours trying to make it so that streaming responses from the LLM is decoupled from the http request that enter the prompt:
https://t.co/vtfXhJIG8h