Nuggt

Instantly go from 0 to 1 on any topic with nuggt! Repo: github.com/shoibloya/nugg… Link: app.nuggt.io made with @firecrawl 🔥

@YoungPhlo_ Hi there, our current implementation uses wizardcoder you can check out its performance on our repo github.com/Nuggt-dev/Nugg… we would also create a video explaining how we got these results!

@OfficialNuggt WizardCoder15B and/or StarCoder

Thinking about starting a video series comparing how different open source models perform as agents... Which model would you like us to test first?

@JulienTech @baseten @StabilityAI Hi Julien, @baseten looks exciting. Will drop you a DM to move this forward thanks!

@OfficialNuggt Hey Nuggt! Would absolutely watch this. Gotta recommend @baseten as a tool to test these models in the video series. @StabilityAI and other ML labs use us to deploy/iterate/log model activity, so I'm thinking it'd be a slam dunk. Can send you free creds to do it all for $0.

If you have any other models in mind just reply here!

@AnxiousFurbii @LOFI911 Interesting, we'll look into this!

@OfficialNuggt @LOFI911 transfer what you're requiring from the user to reason/think about, to the LLM, whenever possible I want "go find 5 experts about machine learning, and tell me about them" and I'm waiting for one of these tools to ask me to choose which of the three top approaches I'll go with

Task: Scrape social media for experts. Prompt: Using {tool:search} to scrape {text:social_media} to find 5 individuals who are experts in {text:topic} using {tool:python}. Using {tool:llm}, generate a short summary of the profiles of the 5 individuals and display it. Day 6/30

@LOFI911 @LOFI911 thanks for the feedback! We're in the process of simplifying the UX for this. Do you think the complexity arises from the prompt's length or its wording?

@OfficialNuggt This is nice but why are the prompts so complicated ?

@WizardLM_AI Look what we did with your model. Amazing work.. would love to collaborate

Thrilled to unveil our Open Source LLM Agent that runs on Wizcoder-15B. Say goodbye to extravagant API fees, pesky security worries, and the constraints of closed models. Watch the Nuggt Agent train a RandomForest on the Titanic dataset and save the ROC Curve in <1 minute.

We're super excited for the future of open source models and equally, the use cases they bring to extend human potential. Check out the GitHub: github.com/Nuggt-dev/Nuggt Join the Discord: discord.com/invite/YZp6jmFr

Task: Generate summaries of an article in another language. Prompt: Using {tool:search}, read the article, {text:article} and generate a 300 word summary using {tool:llm}. Using {tool:python}, translate the summary into {text:language}. Day 5/30

Task: Snake Game Prompt: Print game requirements from {upload:description} using {tool:python}. Based on the requirements build the game in {tool:python}. Save the python class files and main.py under the folder {text:folder} using {tool:python}. Day 4/30

Task: Finding activities to do on a weekend. Prompt: Find {text:number} activities I can do in {text:place} this weekend using {tool:search}. Display the activity name and link to the website. Day 3/30

Task: Find books, podcasts and youtube videos on a given topic. Prompt: For the topic, {text:topic}, give {text:book}, {text:podcast}, {text:youtubeVideo} recommendations using {tool:google}. Display the data in a tabular format using {tool:display} Day 2/30

Task: Create a boiler plate template for a full stack application. Prompt: Using {tool:python} create boilerplate code for a full stack app with frontend in {text:frontend} backend in {text:backend} and database in {text:database}. Day 1/30

[Open Source] Automate any task with a single AI command. We would love to hear your thoughts. You can contribute at: github.com/Nuggt-dev/Nuggt Join Discord: discord.gg/gzdCDM84