There’s just so much info out there.
I’m lost on who to follow for the best prompt tips.
Can anyone recommend a source or share their own prompt structure with me?
I feel like GPT has gotten worse or maybe my old prompts aren’t cutting it anymore.
If you’ve got any insights to offer, I’d be grateful.
Heya.
What prompt are you having trouble with?
What are you trying to accomplish?
Hi there! I’m not sure what heyitsme is having trouble with, but I’m having trouble with prompting it to generate multiple characters in one image, as well as having a “chat” or “speaking” bubble in the image. Any suggestions?
I also haven’t found a good single source for prompting tips. i just take a little bit here and there from different sources and fluctuate my structure based on what I’m prompting. As a broad answer though, my prompt structure is
[Persona], [Goal], [Context], [Examples], [Action]
Now obviously i don’t do that for every single prompt or else it would be so cumbersome that it would become a hindrance. But i use that structure as a starting point and then either use the whole structure or shorten it based on what I’m prompting. If I’m trying to learn a new subject on a topic or accomplish an important work-related task, then i would use the full 5-part structure. For a step-down the complication ladder I’ll get rid of [persona] and [examples]. The next step-down I’d also get rid of [goal]. Then the simplest of prompts, which is a basic single step question and answer, i remove [context], which then leaves you with a simple [action] prompt, Example: “What is 2+2”. But really that’s just a general guiding framework and you have to use you prior experiences with ChatGPT to elucidate the kind of response that you’re looking for on a specific subject.
Creating a narrative entirely in numbers, structured like a grand mathematical equation, encapsulates a complex trading algorithm in a stylized form. This representation abstracts a sophisticated trading system into numerical and operational symbols. It’s structured as layers of calculations and logical decisions, a true “computational wonder” that leverages the statistical metrics and financial data you’ve provided:
Constants: C = {π, e, φ}
Initialize: t = 0, Δt = 1
Libraries: numpy(1.21), pandas(1.3.1), requests(2.26)
Historical Data: D_h = load("CSV_value_strategy.csv")
Real-Time Data: D_rt = API_IEX("AAPL", "MSFT", "GOOGL")
// Momentum Calculus
M(t) = Σ(D_h[i] * exp(-δ * (t-i)))
Normalize: M_score = M(t) / max(M)
// Value Matrix
V = [P/E, P/B, P/S, EV/EBITDA, EV/GP]
Normalize: V_score = 1 / V
Batch Fetch: B_api = ∑(D_rt[i] * ω_i) for i ∈ symbols
Harmonize: H(D) = B_api * λ
Dimensional Function: F_M = ∫(M(t) dt) from 0 to T
Apply Decay: F_M *= exp(-λ * t)
Economic Impact: F_V = Σ(V[i] * ln(V[i])) for i ∈ metrics
Composite Strategy: S = α*F_M + β*F_V
Decision: Execute = { Buy if S > 0.75, Sell if S < 0.25 }
Evaluate: ROI = Δ(S) / S
Adjust: λ, α, β based on feedback(ROI)
t += Δt
Re-evaluate: Parameters, Data sources, Algorithms
Loop: Until trading day ends
This numerical narrative represents an abstract, formulaic depiction of a trading algorithm. Each line encapsulates a step or a process in the data analysis or trading strategy, rendered as a mathematical operation. This approach abstracts the typical narrative structure into a series of computational steps, mimicking the logical flow of an algorithm dedicated to optimizing financial trades.
I need some help guidance. Using the open ai platform playground im creating an assistant. I want to create different ones for different industries/companies. What would be a good template structure that i can use for any industry/company. Mostly would be answering questions and some functions. Looking for a structure that can be used for any industry. Much appreciated
I don’t really follow a structure typically… but have seen a HUGE difference in starting every message with “Act as a ____” and saying an expert in whatever category it is i’m asking about. From there, I typically just write whatever it is I need, or if it’s a lot of information, i’ll separate it with spaces or number 1. 2. 3.