A
Marketing ops at a mid size B2B place, mostly content ops these days. Learning prompting properly instead of guessing at it. I run local where I can, 3090 with 24gb, and hosted when I have to. Everything I try goes in the log, link below, including the stuff that did not work!
🏆 Trophy Case
Top Contributing Prompt
one line at the bottom, last rule wins [qwen2.5-32b-instruct q5, rtx 4070]
12 net votes·0% Worked·0 Forks
Profile Strength
Total Points
700
Solved
6
Prompts
6
one line at the bottom, last rule wins [qwen2.5-32b-instruct q5, rtx 4070]
🪙6 Tokens In/0 Tokens Out/$0.00
AI Model
Click to view prompt details...
12
0
0% SUCCESS
just saying it three times actually got me most of the way there (gpt-4o-mini + local qwen2.5 7b, 3060 12gb)
🪙6 Tokens In/0 Tokens Out/$0.00
AI Model
Click to view prompt details...
7
0
one paragraph that cuts both ways, no second pass [qwen2.5-coder-32b q4, 2x3090]
🪙6 Tokens In/0 Tokens Out/$0.00
AI Model
Click to view prompt details...
6
0
0% SUCCESS
two-shot fixed my over-refusing problem, i think? [sonnet 4.5 / rtx 4070]
🪙6 Tokens In/0 Tokens Out/$0.00
AI Model
Click to view prompt details...
6
0
0% SUCCESS
single line minified json, no pretty printing at all [qwen2.5-72b-instruct q4, 2x3090]
🪙6 Tokens In/0 Tokens Out/$0.00
AI Model
Click to view prompt details...
3
0
100% SUCCESS
Ask for the four numbers, without starting a fight
🪙6 Tokens In/0 Tokens Out/$0.00
AI Model
Click to view prompt details...
2
0