Wow GPT-4V such an eloquent discussion. There are some… minor issues however. `..dxe4` isn't a legal move, since it would involve white going backwards and taking its own piece. Hmm yeah and `Qxg6+` is gonna be tricky because there's no queen on the board. :O
PROMPT ENGINEERING
-
DALL·E 3’s Secret: Dynamic Focus on Image Captions
By
–
The secret behind the ingenuity of DALL·E 3 is its dynamic focus: it emphasizes the image captions. This positions it in a unique space, enabling it to interpret both the image and the narrative directive behind it.
-
Copilot Hits 1M Paid Users With 55% Productivity Gains
By
–
Did you watch the keynote? Numbers are in. 1M paid users for Copilot, and an 55% improvement on productivity for all devs using Copilot. So maybe what you’ve experienced is “shit-in-shit-out” type situation? I don’t mean it’s your fault, but AI is only as good as the user
-
Midjourney prompt for knolling style AI art
By
–
My favorite prompt for Midjourney. Knolling [subject] Knolling robots, AI tools, white background, sketch, –ar 16:9 #midjourneyartwork #MidjourneyAI #midjourneyart #godofprompt #AIArtwork #aiartist #PromptShare
-
SPADE: Automated Test Generation from LLM Prompt Refinements
By
–
Sick of debugging finicky LLM apps? 😫
— LangChain (@LangChain) 8 novembre 2023
Introducing ♠️ SPADE, a tool that automatically generates tests from prompt refinements! 🧪
Paste any prompt from the LangChain Hub and it will tailor eval recommendations using prompt deltas. 🔁
SPADE analyzes version history to suggest… pic.twitter.com/eSmWaPjqc8Sick of debugging finicky LLM apps? Introducing SPADE, a tool that automatically generates tests from prompt refinements! Paste any prompt from the LangChain Hub and it will tailor eval recommendations using prompt deltas. SPADE analyzes version history to suggest
-
GPT-4 Vision and DALL-E 3 iterative avatar generation
By
–
I'm using GPT4 vision and dalle3 together to try and reproduce my abstract avatar. I asked it to: – analyze the original image
– analyze the generation
– iterate the prompt to improve accuracy These were the results of 4 iterations. -
Generate test cases automatically with LLM prompt engineering
By
–
You can get arbitrarily creative with it, too: cat script.rb | llm -s ‘This code is janky. Write some test cases for it, focusing particularly on validating the command line args and that it exits prior to calling the API if it detects an error.’ > test/script_test.rb
-
Using DALL-E with detailed prompts for better image generation options
By
–
Dall-E, though I usually use Midjourney. I did something a bit different than my usual one-line prompt on the title/themes/style to hit and asked it to give me some options, then said:
-
Managing LLM Context Windows: Input Truncation Strategy
By
–
"Why did you pipe the output of the git commands to head above?" Because LLMs have limited context windows and if you exceed the length of them this tool fails gracefully, so by habit I chop the input to it down prior to asking to process it.
-
Using Git diffs with LLM to summarize commit changes
By
–
git show 57ee | head -c 4000 | llm -s 'This is the first part of a diff. What did this commit actually *do*? Your answer should fit in a tweet.'