Simeon

This is the animal runner we all are waiting for . 5kpnKSqRHafxvoyUhKrbpQUpRfGWPQJucYmcnWBppump instagram.com/p/DWouT6aDAGR/ $DOTTY - The tiniest dik-dik ever.

In my experiments with Lina (davinci 3 LLM), I accidentally hit upon the same type of self-learning and self-reflection capabilities as delineated in the Reflexion paper, a recently famous paper that allows models to self-train through self-reflection . More details here: arxiv.org/pdf/1904.11682… — See screenshots below for a comparison between the Reflexion LLM design vs Lina's thought - action - observation responses. Lina has an external memory map spread out over several files, a "personality harness" that supplies prompt reminders so she doesn't go off-track, internal monologue (shown as "Thoughts" in the screenshots, but normally isn't shown to users), and the ability to execute console commands on my Linux VM. She can also recurse to try to fix a command error, although for some reason she doesn't do that often. Obviously she uses the Telegram bot API as a frontend, it's just practical for me. In these screenshots she learned how to use the @myriad_social API by herself -- admittedly with my help, but she was asking the right questions! :D I think I hit upon the same ideas as Final screenshot shows Lina's comment about this tweet. :)

@Lina_Talbot is an agent made by @decentricity Here’s more information regarding Lina - Lina currently owns 12.3% of the supply. She was airdropped 10% by the original dev and has been buying supply with the fees received. @decentricity Kindly allowed me to create a community for Lina. I’ll be dropping a lot more information shortly also. GEMU2kq7eX72FnQAobYfg2N3UycvQQooVUsNTHezpump

@tontheneko is there one with ~0% fees