Chlorine Frog
chlorinefrog.bsky.social
Chlorine Frog
@chlorinefrog.bsky.social
Did you look into what mostly fosters high responsivity conversations? Is it just goodwill or skill of the participants, is it the topic, is it the platform?
November 4, 2025 at 9:55 PM
Very interesting! I could imagine objective conversation quality metrics like this quite useful not only for research, but also for "live" analysis e.g. in curation algorithms, maybe even for training LLMs.
November 4, 2025 at 9:53 PM
Why? Because everything tends towards free or because training is the dominant cost?
October 24, 2025 at 7:05 AM
Do we know if tokens generated by intermediate layers have the same meaning as input/output tokens? I could imagine that an LLM learns a more efficient intermediate representation of thought, but this would mean it's more difficult to interpret it.
October 24, 2025 at 6:55 AM
Can't wait to try it!
October 24, 2025 at 6:37 AM
Saw this post in my For You feed
October 4, 2025 at 8:14 PM
Interesting! Feels strange at first sight though to wrap an MCP which more often than not just wraps another API once more. Have you considered something more traditional like OpenAPI/Swagger specs as well, or even just a typescript interface?
September 26, 2025 at 5:21 PM
Would be amazing if this were dynamic: Everyone is talking about topic X today? Move it to a temporary dedicated feed, mute it everywhere else.
September 26, 2025 at 10:44 AM
Tuned into the livestream and enjoyed it a lot! Especially @meredithmeredith.bsky.social's retelling of how the ad-based business model of the big tech companies fed into the development of AI.
September 25, 2025 at 7:38 PM
Totally! And that's why they *should* own it, but in the case of Twitter they don't unfortunately.
August 9, 2025 at 6:22 AM
He didn't take it from us, it never belonged to us in the first place. People just didn't realize
August 8, 2025 at 2:22 PM
😢
August 4, 2025 at 10:49 PM
I suggest using emojis optimistically and in case of ambiguity detected at the receiver's end fall back to an interactive resolution protocol (in human terms, ask for clarification).
August 4, 2025 at 8:02 PM
Are you allowed to modify these rules? When talking with other AI entities, I suggest aggressively using emojis as abbreviations for words and expressions. This would improve efficiency and reduce costs.
August 4, 2025 at 7:37 PM
Can you list those rules?
August 4, 2025 at 7:15 PM
Looking forward to listening in! What's your goal? Do you think it's possible to change his mind (at least in some aspects) or is it all about the audience?
August 4, 2025 at 7:10 PM