25/05/2026 2:25 pm
Topic starter
Notifications
Clear all
DeepSeek Forum
2
Posts
4
Users
0
Reactions
7
Views
0
I'm trying to get DeepSeek V4 Flash to actually write decent prose for a sci-fi novella I'm entering into a local contest next month. I've looked at those mega prompts on GitHub but they seem way too bloated for a fast model like Flash and some guys on Discord were saying XML tags help while others say just keep it simple. Im just confused if I should be feeding it specific style guidelines or if that makes it more robotic.
Heres what I need:
- No purple prose or tapestry words
- Strong character voices
- Fast output since I'm on a tight timeline
- Sticking to the cheap API tier for now
Should I use a persona prompt or just list out negative constraints like dont use cliches?
11
25/05/2026 2:25 pm
Unfortunately, my experience with DeepSeek V3 API showed simple negative lists were disappointing.
- XML tags: Strongest adherence.
- Persona: Often results in cliches. Use XML to keep character voices consistent.
11
25/05/2026 2:40 pm
Stumbled upon this because I was also nervous about wasting credits. Using TypingMind Desktop App helped me test small prompts safely. Tbh, just giving the model one solid paragraph describing the tone's a decent option. It works better than those massive GitHub prompts that'll make the prose feel stiff. Sticking to simple instructions and setting temperature to 0.7 kept everything fast for me.
