Practical overview
When it fits
Use it when an ML task can be expressed with a dataset, a goal and a measurable evaluation metric, and you want an agent to explore successive code revisions.
Who it is for
- Machine-learning practitioners prototyping pipelines
- Agent-architecture researchers testing search heuristics, evaluators or LLM back ends
- Engineer-researchers replicating or extending the underlying approach
Documented uses
- Generate machine-learning code from a natural-language task description
- Iteratively debug and improve an ML solution using metric feedback
- Benchmark alternative code paths and inspect the resulting search tree
- Prototype experiments through a CLI or Streamlit interface
Requirements & limits
The repository is public and released under the MIT license. The documented quick-start workflow uses Python 3.10 or later and an LLM API key.
- A user-defined metric is central to the search and improvement loop.
- The project offers both CLI and Streamlit workflows.
- The implementation is Python-based and includes visualisation and configuration utilities.
- The model integration is described as supporting OpenAI, Anthropic, Gemini and local models compatible with the OpenAI API.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Run a documented AIDE ML optimisation
Install AIDE ML and run an optimisation against a dataset, goal and evaluation metric.
Before you start
- Python 3.10 or later
- An LLM API key, shown in the quick-start example as OPENAI_API_KEY
- A dataset directory, a goal and an evaluation metric
Steps
Install or upgrade the package with pip install -U aideml.
Set the LLM key in the environment.
Run the aide command with data_dir, goal and eval values.
Inspect the generated best_solution.py and tree_plot.html files in the run log directory.
Expected outcome
AIDE ML produces a best code solution and an HTML view of the searched solution tree.
Guide limitations
- The documented command uses an LLM API key.
- The result depends on the supplied dataset, goal and metric.
Launch the AIDE ML Streamlit interface
Set up the repository's documented web UI for running AIDE ML interactively.
Before you start
- A local clone of the AIDE ML repository
- An LLM API key
- Data to upload, plus a goal and metric
Steps
Clone the repository and change into its directory.
Install the project in editable mode with pip install -e .
Change into aide/webui and run streamlit run app.py.
Use the sidebar to provide the API key, upload data, set the goal and metric, then press Run AIDE.
Expected outcome
The UI displays live logs, the solution tree and the best code.
Guide limitations
- The documented UI workflow is local and depends on Streamlit.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 30 September 2026
- Original website title
- WecoAI/aideml · Official homepage title · retained in its source language