Aquarium @aquariumlearn

Aquarium helps teams streamline the process of finding issues, validating fixes, and adding data for their machine learning datasets. aquariumlearning.com Joined April 2020

Tweets

70
Followers

253
Following

1
Likes

5

Peter Gao @pgaooo

5 days ago

Launching Tidepool Warehouse Native! Now you can get started with analysis of your text within minutes and automatically sync Tidepool’s outputs back to your data warehouse, all without storing sensitive data outside of your cloud environment. tidepool.so/blog/launching…

0 1 1 120 0

Peter Gao @pgaooo

a month ago

Providing examples is a useful tool for prompting LLMs, but did you know that the specific choice of examples has a dramatic impact on performance? Our own Lucy Cheng made a guide on how to pick examples to get the highest accuracy... tidepool.so/blog/picking-t…

0 1 1 113 0

Peter Gao @pgaooo

a month ago

We made a new walkthrough video for Tidepool! Tidepool is an analytics tool for large text datasets. Categorize and surface patterns in unstructured text data like user prompts to LLMs, user feedback, social media posts, and more. youtu.be/2yGTZBAH1T4?si…

0 1 0 124 2

Peter Gao @pgaooo

2 months ago

We just launched our new website for Tidepool! Tidepool is a data analytics tool for text data. It imposes structure on unstructured text so you can derive insights from text with your existing BI stack. Ping me if you want to try it! tidepool.so

1 2 5 470 1

Peter Gao @pgaooo

2 months ago

LLMs are enabling new ways to design UIs! Charlene walks through some of the most impactful examples of how LLMs are changing the ways we interact with computing. tidepool.so/2024/02/21/eme…

0 1 2 107 1

Peter Gao @pgaooo

2 months ago

Tidepool lets you derive some useful business insights from text data you already have! I wrote a walkthrough of the Yelp Review dataset to show how you can understand attributes of text data and their correlation to a metric of interest. tidepool.so/2024/02/14/ana…

0 1 2 106 0

Peter Gao @pgaooo

3 months ago

We've learned a lot of neat tricks while prompt engineering LLMs for Tidepool. To share some of these learnings, we wrote a guide on how to do JSON chain-of-thought prompting with GPT-4! tidepool.so/2024/02/08/cha…

0 1 1 168 1

Charlene Chambliss @blissfulchar

3 months ago

A lot of friends have asked me about my process for using LLMs to learn more quickly. Well, I finally wrote a post on the Tidepool blog about it! These are a few of my top strategies for efficient learning with LLMs: tidepool.so/2024/01/22/usi…

0 2 1 153 0

Peter Gao @pgaooo

3 months ago

LLM engineering can seem like dark magic, but making some simple changes to your prompts can greatly improve your RAG quality! tidepool.so/2024/01/17/don…

0 2 4 117 0

Peter Gao @pgaooo

4 months ago

We wrote a guide talking about how to diagnose and fix common issues with RAG systems! tidepool.so/2024/01/03/deb…

0 1 1 94 0

Peter Gao @pgaooo

6 months ago

I know a lot of engineers who used to work in autonomy who are now working on LLMs. And it makes sense, too - it turns out a lot of lessons about improving robotic systems applies to improving generative AI systems! tidepool.so/2023/11/08/how…

0 1 3 187 1

Peter Gao @pgaooo

6 months ago

Tidepool's Quality Reports help you find where your model is failing in production and break down your top failure cases by root cause. Prioritize work with the most impact and improve your LLM app faster! loom.com/share/5d333a24…

0 1 1 132 0

Peter Gao @pgaooo

7 months ago

A common question LLM devs have is, "is my LLM doing what I want it to?" We wrote an article talking about some easy tips to make sure that your model is doing what you want it to before and after you ship it to users! tidepool.so/2023/10/03/is-…

0 1 2 199 2

Peter Gao @pgaooo

7 months ago

Tidepool allows you to understand text interactions between your users and your LLM apps. Figure out common patterns in how people use your product and understand areas where your model is performing well / badly so you can make your product better! loom.com/share/94eb15a8…

0 1 2 185 0

Peter Gao @pgaooo

8 months ago

A very common use case for LLM applications is to translate natural language queries into structured outputs like DSL's or SQL queries. We wrote a guide on how to implement this for your own app! tidepool.so/2023/09/13/con…

0 1 3 252 0

Peter Gao @pgaooo

8 months ago

It's tempting to look at the noise and the hype in the LLM space and get discouraged. The reality is that most LLM applications will not work, but a minority of them will turn out to be super valuable! tidepool.so/2023/09/06/mos…

0 1 2 227 0

Peter Gao @pgaooo

9 months ago

Fine-tuning introduces a lot of complexity and negates one of the most important advantages of LLMs, which is the ability to do few or zero shot learning. There's a lot of ways to get LLMs working on your own data that isn't so difficult. tidepool.so/2023/08/17/why…

1 1 1 264 1

Peter Gao @pgaooo

9 months ago

AI text interfaces unlock a new way for people to interact with software, yet traditional product analytics tools don't work well on user text interactions being generated by these LLM apps. Can we build something better? tidepool.so/2023/08/01/how…

1 1 1 174 0

Aquarium @aquariumlearn

9 months ago

So many PM / design / engineering folks are scrambling because their leadership asked them to "figure out what we can do with AI." To help, we made some inspo about the most successful product applications of LLMs we've seen so far! tidepool.so/2023/07/26/5-t…

0 2 2 103 0

Aquarium @aquariumlearn

9 months ago

We're live on Product Hunt! producthunt.com/posts/tidepool…

0 0 3 91 0

Make games with @rosebud_ai : Prompt, Publish, Play. 🌹 https://t.co/zDq6c6v7D1 🎓 Founder. PhD AI Berkeley 👼 Angel investing https://t.co/VMlOAnTC1j

0 0 4 20 0

Comet @Cometml

3 years ago

Today in Heartbeat: @DN06_90 uses Comet & @aquariumlearn to build an evaluation system for debugging an #ObjectDetection model that focuses on fixing problems in the underlying data, instead of the model. Read the full tutorial here: bit.ly/3mOgzQ2 #mlops #deeplearning