pplonski86

James Clear, Atomic habits. I really enjoy it.

Backend in Django Rest Framework. Postgres as db. Frontend Typescript and React. Deployment with docker compose on AWS.

How do you reuse your history? Is it used only in Obsidian? Is there any option to load it as context to ChatGPT?

YAML is perfect for simple scenarios. But users produces with it really complex use cases.

Is it possible to write Python package that based on YAML specification produces Python API? User will code in Python and YAML will be the output.

I was working on YAML syntax for creating UI. I converted it to Python API and Im happy. For exmple, dynamic widgets in YAML were hard, in Python they are strightforward.

Before learning about RAG I thought that it is recurrent LLM agent that traverse over documents. After some study I must say that VectorDBs are boring.

It is quite similar to https://github.com/KillianLucas/open-interpreter

Such terminal tools are quite like AI agents, it should think about steps and then execute one by one.

What are other than fine-tuning methods to make LLM smarter? Im familair with RAG - Retrival Augumented Generation.

Congratulations on launch! I've seen your tweet about unlimited context for LLMs :)

Do you have option to manually provide text that will be used as LLM memory?

Can it be easier to do RAG? Do we always need to have Vector DB? Why LLM can't do search through the context by itself?

Have you used any benchmark to test agents? I'm currently looking for REST API usage benchmark for LLMs.

Nice analogy - yes, something like this. What is more, LLM notes can be hierarchical, to have some kind of generalization.

It's just example, I've manually created it. But I think LLM can do such memory notes for itself and include as context.

I think this can be easily fixed, if LLM can do notes on what's going on. If it will have additional context before the prompt:

```

You are home assistant. Here is information what's going on in the house:

It's 4PM. Bob likes Chopin Fantaisie-Impromptu. Alice likes Mozart Rondo in D. Bob is in the house. Alice will be back from office at 5PM.

You get a prompt: I'm tired, play my favorite

```

For the above input any LLM will play Chopin.

The service unfetch.com looks cool. Can I run it locally on my machine, not in the cloud? Is it open-source?

I've seen Noteable+ChatGPT demo, where user can chat with ChatGPT and responses where executed in the Noteable-hosted Python notebook. It was cool!

It would be also cool to have such plugin for Google Colab.

I hope someone will come with a new way to interact with LLM models other than chat UI. It would make code writing even more faster.

How RestGPT differs from ToolLLM or Gorilla?

papers:

1. ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs https://arxiv.org/abs/2307.16789

2. Gorilla: Large Language Model Connected with Massive APIs https://arxiv.org/abs/2305.15334

I did some internet search and the website is open source https://github.com/Idered/chalk.ist

And it looks you are from Białystok - greetings from Łapy!

Great website! Congrats.

1. Any chance to make it open-source?

2. Is there an option to build animation of how code is written? Export code animation as GIF?

Could you please describe how it works?

I was playing with their API for half day for code generation and the cost was few dimes. I was writing package for making matplotlib plots. https://github.com/mljar/plotai

Are you building some public service that heavily relay on OpenAI API?

Do you have any demo or github with your prototype? Ive checked your blog but no recent posts.

I was also thinking about deeper notebook and LLM integration for linear scripts.

Congrats. Is there any way to provide feedback for LLM? Is there any way to ask for approval before code execution?

I wrote a simple wrapper around Matplotlib and ChatGPT-3.5-turbo. The LLM response is a Python code that is executed to get charts. It is working very nice. Here is a repo https://github.com/mljar/plotai - you will find two videos in the readme. Maybe you should work on your prompts?

I was searching for similar article. Im working on AutoML python package where I use different packages to train ML models on tabular data. Very often the memory is not properly released by external packages so the only way to manage memeory is to execute training in separate processes.

Hey! Thank you. I appreciate you response! That was wild run. Emotional roller coaster. Anyway, I made mljar AutoML open source, it is available in our github https://github.com/mljar

I needed multi-line string, because I know that there will be several iterations of prompt creation.

The LLM has access to user prompt, 5 first rows of DataFrame and template prompt (for example with sentence 'Output only Python code in response').

I was very surprised with Australia example! The LLM has access to only first 5 rows of the DataFrame and Australia is not there. The LLM was able to produce the correct code, that genralizes well on full DataFrame.

The easiest way to create plots in Python and Matplotlib. The plotai is using LLM to generate code and plots.

The idea:

1. User provides input DataFrame and prompt.

2. The PlotAI constructs a prompt for LLM, which contains the first 5 rows of DataFrame and the user's prompt and asks for Python code as output.

3. Returned Python code is executed, and the plot is displayed.

The simplest possible API for plotting:

  # import packages
  import pandas as pd
  from plotai import PlotAI
  # create some data
  df = pd.DataFrame({"x":[1,2,3], "y": [4,5,6]})
  # do a plot
  plot = PlotAI(df)
  plot.make("scatter plot")

The PlotAI class has only one method, make().

It works in Python scripts and in notebooks (Jupyter, Colab, VS Code).

Here is benchmark done by independent team of researchers https://openml.github.io/automlbenchmark/

I think most of overfitting is avoided with early stoppoing technique.

The underfitting can be avoidwd with using large training time.