I believe there is a big opportunity for LLM guardrails due to the non-
deterministic nature of the Transformer architecture.
However, the just announced Claude Cowork still warns humans to stay in control: https:// claude.com/blog/cowork-research-preview
I assume this is because their non-human guardrails are not good enough yet to
fully validate the output of an LLM.
What non-human guardrails does Axonflow employ to enforce a policy rule with X% confidence on a prompt / LLM output?
Personally I really enjoy doing analog photography. As the feedback loop is very long, technique is important. Also, since the number of pictures is limited, I always have to ask myself the question: is there actually a picture here?
I usually consider programming books as a reference and don't "read" them. My way of getting value out of these books is remembering the high-level contents and diving deeper when there is the need (e.g, a coding problem that I need to take care of)
Great idea! Did you deploy a speech-to-text pipeline to achieve this? I always thought it would be relatively expensive to do podcast-to-text translation at scale (compared to the gains) but maybe I just didn't optimize it well enough :)
It is very likely to be victim of survivorship bias here. It's not that there were no bad houses back in the day, they are just not here anymore. So it is easy to believe people built better houses in the past.
Writer here. As someone who spends a lot of time in public transport I am always disappointed in current public transport planners, since they are ‘static’ and can’t handle well disruptions.
I created the public transport assistant concept to explore the possible advantages of live location data within public transport and to investigate the possibility of creating a more dynamic route planner. For example, once we know in which train a passenger is, we can check when this train will arrive at a certain station and act upon it.
My next step will be to create a minimum viable product that encapsulates a basic route planning system (e.g. OpenTripplanner) and the described vehicle classification system.
Let me know what you think, feedback is always welcome!
Shameless self plug here, few months back I created a project https://www.contenttagger.com that gives you hints for Schema tags based on a parent Schema attribute. You can also validate it with Google devtools + Facebook graph.
ContentTagger automatically shows you the 'child' attributes from a chosen domain or range attribute so it becomes more easy to select something. On top of that I'm able to pass a link with the code to Google Structured Data Developer console and Facebook Graph, so easy testing!
In the future I want to add more schemas (currently only schema.org) and create a library which you can use to query linked data yourself with SparQL.
Stack: Node, MongoDB, React
If you have any questions or feedback let me know so I can make the product even better, thank you!
However, the just announced Claude Cowork still warns humans to stay in control: https:// claude.com/blog/cowork-research-preview I assume this is because their non-human guardrails are not good enough yet to fully validate the output of an LLM.
What non-human guardrails does Axonflow employ to enforce a policy rule with X% confidence on a prompt / LLM output?