datadm

We've made a lot of data tooling things based on LLMs, and are in the process of rebranding and launching our main product. 1. sketch (in notebook, ai for pandas) https://github.com/approximatelabs/sketch 2. datadm (open source, "chat with data", with support for the open source LLMs (https://github.com/approximatelabs/datadm) 3. Our main product: julyp. https://julyp.com/ (currently under very active rebrand and cleanup) -- but a "chat with data" style app, with a lot of specialized features. I'm also streaming me using it (and sometimes building it) every weekday on twitch to solve misc data problems (https://www.twitch.tv/bluecoconut) For your next question, about the stack and deploy: We're using all sorts of different stacks and tooling. We made our own tooling at one point (https://github.com/approximatelabs/lambdaprompt/), but have

From making a few variations on data chatbots in the past year, I found that my favorite / most fun to use ones seem to be more "chain-of-thought" and conversational rather than "retrieval-augmented" style. Less about one-shotting the answer, and more about showing its work, if it errors, letting it self-correct. Latency goes up, but quality of the entire conversation also goes up, and feels like it builds more trust with the user. Key steps are asking it to "check its work", and watching it work through new code etc. (I open-sourced one version of this: https://github.com/approximatelabs/datadm that can be run entirely locally / privately) From their article: I'm surprised they got something working well by going through an intermediate DSL -- thats moving even further away from the source-material that the LLMs are trained on, so it's a

Pretty good examples and simple explanations. I didn't realize Claude 2 was so good at working with PDFs natively. I wonder if they're doing anything special? Is this just due to larger context length they have? Also, biased opinion on my part: I'm especially interested in watching how these things affect data science and data literacy as a whole. Code interpreter is a game changer in my opinion, the most powerful tool that I think deserves all the press it is getting. Also: I released an open source code-interpreter for data (https://github.com/approximatelabs/datadm) and even though I know how to code and use Jupyter daily, I still find myself doing analysis with it instead. All in all, it does seem like the different models and agents are gaining "specialization" skill is actually good for the user (rather than just using a single jack

I completely agree that greatly increasing data accessibility is a huge unlock and value add. A package I open sourced recently might be useful for use cases like this, https://github.com/approximatelabs/datadm It's essentially a chatGPT code interpreter, specifically designed to work with data, that can be run entirely on open models (eg. StarChat). True local mode operation.