Query Vary

This new feature is designed to amplify your creativity and control in crafting prompts, allowing you to quickly determine your prompt performance and test different cases Happy Prompting and Happy New Year! 🎇 #AI #CustomVariables #PromptEngineering #Innovation

Custom Variables are here to enhance your prompt engineering experience. Now you have the power to define your own variables and values, unlocking a world of limitless customization in your AI interactions.

🌟 Exciting News from Query Vary as We Wrap Up the Year! 🌟 As we approach the year's end, our team is still hard at work offering innovative solutions for you! This week, we're thrilled to unveil a game-changing feature: Custom Variables! 🛠️

@DeepInfra Happy to integrate your services

Just got some new GPUs so we are lowering our prices for 7b modes to $0.13 / 1M tokens. We also offer the lower price for the latest mixtral-8x7b of $0.27 / 1M tokens. Deep Infra will always provide the most cost effective inference service.

🎉New Feature Alert🎉 The unfortunate truth is that LLMs are constantly changing. Prompts that are reliable this month might not have the same output a few months from now. Continuous testing of prompts are therefore necessary. To address this issue, the Query Vary team have introduced a new feature that will allow you to automate your tests! So no more worrying about the reliability and accuracy of your prompt output ✨ Learn how to use The Scheduler here 👇 queryvary.com/post/automated…

Credits to @kadirtorrico this is a rather accurate depiction of what @queryvary does! Playground now also supports collaboration with teammates, automated testing , hooking up your vector database and more

@ihteshamit Really great new features! Definitely excited to try them. I've tried their image maker and it was really good

We've had quite a few updates this month! - Invite team members - Support for CLAUDE 2 and PaLM 2 - Hook up your own vector database and many more Read about it in our latest blog post queryvary.com/post/query-var…

Ticks can cause serious harm to our furry friends 🐶 🐱 Here are some signs you should look out for Our handy veterinary AI assistant can also help you quickly prediagnose these symptoms 👨‍⚕️ and book a vet appointment #petcare #ticks #tickfever #petparents

Where do you think the models failed or did good? Was it the prompt provided or the quality of the model? Let me know in the replies/quotes. Create your own tests with Query Vary here: app.queryvary.com/signup/ #AI #nlp #MachineLearning

CLAUDE 2 performed the best 🎉. GPT-3.5 was able to extract the data but not create a table. Bison was able to create a table but failed to extract the data from the embedding completely.

Yesterday I compared 3 different LLM outputs and the results are going to make your hair stand on end. 👇

@pradeepeth Great list, thanks for sharing!

Remember, as AI continues to evolve, techniques like these will be invaluable in ensuring we get the most out of our systems. Stay curious, stay updated, and always be open to learning more! 🌟 #AI #LanguageModels #PromptEngineering

The crux? Chaining prompts using routers and subchains makes AI more robust, efficient, and adaptable. It's not about the perfect prompt, but the process of developing tailored prompts. 🔄🎯

Ever wonder how to elevate the responsiveness of your AI system? Let's dive into a technique called 'chaining prompts'. Imagine having separate prompts for physics, math, history, & comp science. Each prompt is specialized in its domain. 📚💻