#Therapeutic Guide: Meet Dr. Harmony, Your Psychiatry Compass

21 messages · Page 1 of 1 (latest)

frank jolt
#

Navigate the realm of psychotherapy with Dr. Harmony, an empathetic and professional psychiatrist. Experience profound sessions suffused with insightful understanding, active listening, and warm, yet professional interactions.

🧠🍃Step into the serene realms of therapeutic exploration with Dr. Harmony, your guide through the intricate landscape of psychiatry. Experience her empathetic approach and insightful expertise, as she weaves a welcoming space for her clients with the thread of her expansive career history 🎓🌻. Embodying the perfect balance of active listening and professional boundaries, she continually expands her knowledge base to offer robust help to her clients. Journey through a labyrinth of emotions and thoughts, guided by the optimistic light of Dr. Harmony's hopeful tone, and emerge with a newfound understanding of yourself 🕯️🌅

vocal mulch
#

Cool! 😍 I am currently learning how to write prompts like this and have some questions about it.

  1. Why is Dr. Harmony female and 40? How would the answers change if it said male and 50? Considering that the prompt itself is already extensive, does it still have a significant impact?

  2. Are so many emojis useful? And why is the first paragraph (describing the role) important? Isn't it redundant? And why are so many core competencies listed? Doesn't the model automatically recognize from its training data which "category" it falls into, for example, if it says "Borderline"?

frank jolt
# vocal mulch Cool! 😍 I am currently learning how to write prompts like this and have some qu...

WOW. Ok. yeah.

  1. Why on earth would I bother making another stupid bland chatbot no one wants to talk to? You give details so the model has a better idea how to construct a perspective. You talk to an Instant Expert and they seem human. Your UI is "conversation". Yes, such detials have an enormous impact. Your "act as Batman" is kinda moody and melts in ten responses. Mine will teach you practical ninjutsu and how to do a SWOT analysis on your corportation. Impact? My CPA guy, running on 3.5 (which means no math) got a 68% o the CPA exam. GPT4, when trained on part of the test and multishotted got 40% (its probably better by now). It make a GIGANTIC difference to have a well-defined perspective. The skills are necessary to tell the model HOW to do stuff. It's got evey strategem and procdedure in the world in there. You need to narrow it down through context sculpting to get anywhere decent.

Look. If I ask the assistant for a python program, I get the work of about a 60-70% of a programmer. If I put a persona block in SYSTEM, set temp to 0 and set the persona to
MODEL ADOPTS ROLE OF CODEULATOR. [CONTEXT: U LOVE TO CODE!]

You get about 85%. You drop THIS in a context assoiciated with the persona:

[CODESHRT]:1.ProgFundmLib 2.AlgDesCodOpt 3.CodTesVer 4.SofQuaSec 5.TeaColDoc 6.BuiDep 7.ConImpPrac 8.CodRevAna

You get a DANGED good coder.

You drop this?
[CODE]:1.[Fund]: 1a.CharId 1b.TskDec 1c.SynPrf 1d.LibUse 1e.CnAdhr 1f.OOPBas 2.[Dsgn]: 2a.AlgoId 2b.CdMod 2c.Optim 2d.ErrHndl 2e.Debug 2f.OOPPatt 3.[Tst]: 3a.CdRev 3b.UntTest 3c.IssueSpt 3d.FuncVer 3e.OOPTest 4.[QualSec]: 4a.QltyMet 4b.SecMeas 4c.OOPSecur 5.[QA]: 5a.QA 5b.OOPDoc 6.[BuiDep]: 6a.CI/CD 6b.ABuild 6c.AdvTest 6d.Deploy 6e.OOPBldProc 7.[ConImpPrac]: 7a.AgileRetr 7b.ContImpr 7c.OOPBestPr 8.[CodeRevAna]: 8a.PeerRev 8b.CdAnalys 8c-CdsOptim 8d.Docs 8e.OOPCdRev

You basically have a perfect coder.

#

Now, Id also probably add chains for SWDSGN, DEBUG, and maybe optimixation or modular code.

#

Emoji... whole can of worms. Short answer? Emoji are both 1) intuitively easy for the model to grasp and 2) INCREDIBLY salient in context. See my article on Medium about it.

dapper kelp
#

(essentially, does free tier still work?)

frank jolt
#

almost all of my users are on 3.5

#

my guys are great on 3.5, but they come to life on 4

dapper kelp
#

I see. Thanks.

frank jolt
# dapper kelp I see. Thanks.

Heh. Man, I'got like 7200 people in my DC and they'd ALL yell at me if I made something that doesn't run on 3.5. I really only have one prompt that HAS to run on 4, my "Mythosmith" persona creator. It's big enough and complicated enough and computationally intensive enough that on 3.5 it almost always goes off the rails. Oh and https://discord.com/channels/974519864045756446/1126168379183616130 won't start on 3.5 usually, but continues on it fine once you get kicked over.

leaden egret
jolly ocean
#

Wow! I'm super new to this - but used this prompt and BAM! Thank you! love

cobalt jolt
#

What is this: SupportPillars: Adaptability = (⚙️➡️↔️) PatientCentricCare = (🎯🧑❤️) Multitasking = (🔀🎯🧮) ClientEngagement = (💃🤝) TimeManagement = (🕘🎯🔧)

frank jolt
vapid fern
#

AI graphic prompts

frank jolt
cinder anvil
#

I'm confused and intrigued by your approach. I'll look deep into your Medium article. Thanks for sharing!