| Name | autogluon.assistant JSON |
| Version |
0.1.0
JSON |
| download |
| home_page | None |
| Summary | ML Assistant for Competitive Machine Learning |
| upload_time | 2024-11-19 23:27:29 |
| maintainer | None |
| docs_url | None |
| author | AutoGluon Community |
| requires_python | <3.12,>=3.8 |
| license | None |
| keywords |
|
| VCS |
 |
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
<div align="left">
<img src="https://user-images.githubusercontent.com/16392542/77208906-224aa500-6aba-11ea-96bd-e81806074030.png" width="350">
</div>
# AutoGluon Assistant
[](https://pypi.org/project/autogluon.assistant/)
[](./LICENSE)
[](https://github.com/autogluon/autogluon-assistant/actions/workflows/continuous_integration.yml)
AutoGluon Assistant (AG-A) provides users a simple interface where they can input their data, describe their problem, and receive a highly accurate and competitive ML solution — without writing any code. By leveraging the state-of-the-art AutoML capabilities of [AutoGluon](https://github.com/autogluon/autogluon) and integrating them with a Large Language Model (LLM), AG-A automates the entire data science pipeline. AG-A takes [AutoGluon](https://github.com/autogluon/autogluon)'s automation from three lines of code to zero, enabling users to solve new supervised learning tabular problems using only natural language descriptions.
## 💾 Installation
AutoGluon Assistant is supported on Python 3.8 - 3.11 and is available on Linux, MacOS, and Windows.
You can install with:
```bash
pip install autogluon.assistant
```
You can also install from source:
```bash
git clone https://github.com/autogluon/autogluon-assistant.git
cd autogluon-assistant && pip install -e "."
```
#### Beta Features
AG-A now supports automatic feature generation as part of its beta features. To enable these features, please install the beta version dependencies using the following command:
```bash
pip install -r requirements.txt
```
### API Keys
#### Configuring LLMs
AG-A supports using both AWS Bedrock and OpenAI as LLM model providers. You will need to set up API keys for the respective provider you choose. By default, AG-A uses AWS Bedrock for its language models.
#### AWS Bedrock Setup
AG-A integrates with AWS Bedrock by default. To use AWS Bedrock, you will need to configure your AWS credentials and region settings:
```bash
export AWS_DEFAULT_REGION="<your-region>"
export AWS_ACCESS_KEY_ID="<your-access-key>"
export AWS_SECRET_ACCESS_KEY="<your-secret-key>"
```
Ensure you have an active AWS account and appropriate permissions set up for using Bedrock models. You can manage your AWS credentials through the AWS Management Console. See [Bedrock supported AWS regions](https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-regions.html)
#### OpenAI Setup
To use OpenAI, you'll need to set your OpenAI API key as an environment variable:
```bash
export OPENAI_API_KEY="sk-..."
```
You can sign up for an OpenAI account [here](https://platform.openai.com/) and manage your API keys [here](https://platform.openai.com/account/api-keys).
Important: Free-tier OpenAI accounts may be subject to rate limits, which could affect AG-A's performance. We recommend using a paid OpenAI API key for seamless functionality.
## Usage
We support two ways of using AutoGluon Assistant: WebUI and CLI.
### Web UI
AutoGluon Assistant Web UI allows users to leverage the capabilities of AG-A through an intuitive web interface.
The web UI enables users to upload datasets, configure AG-A runs with customized settings, preview data, monitor execution progress, view and download results, and supports secure, isolated sessions for concurrent users.
#### To run the AG-A Web UI:
```bash
aga ui
# OR
# Launch Web-UI on specific port e.g. 8888
aga ui --port 8888
```
AG-A Web UI should now be accessible in your web browser at `http://localhost:8501` or the specified port.
### CLI
Before launching AG-A CLI, prepare your data files in the following structure:
```
└── data # Data files directory
├── train.[ext] # Training dataset (required)
├── test.[ext] # Test dataset (required)
└── description.txt # Dataset and task description (recommended)
```
Note:
- The training and test files can be in any tabular data format (e.g., csv, parquet, xlsx)
- While there are no strict naming requirements, we recommend using clear, descriptive filenames
- The description file is optional but recommended for better model selection and optimization. It can include:
- Dataset description
- Problem context
- Evaluation metrics
- Any other relevant information
Now you can launch the AutoGluon Assistant run using the following command:
```bash
aga run [NAME_OF_DATA_DIR] --presets [PRESET_QUALITY]
# e.g. aga run ./toy_data --presets best_quality
```
We support three presets, including `medium_quality`, `high_quality` and `best_quality`. We use `best_quality` as a default setting.
After the run is complete, model predictions on test dataset are saved into the `aga-output-<timestamp>.csv` file. It will be formatted according to optional `sample_submission.csv` file if provided.
#### Overriding Configs
You can override specific settings in the YAML configuration defined in the [config folder](https://github.com/autogluon/autogluon-assistant/tree/main/src/autogluon/assistant/configs) using
the `config_overrides` parameter with format `"key1=value1, key2.nested=value2"` from the command line.
Here are some example commands on using configuration overrides:
```bash
aga run toy_data --config_overrides "feature_transformers.enabled_models=None, time_limit=3600"
# OR
aga run toy_data --config_overrides "feature_transformers.enabled_models=None" --config_overrides "time_limit=3600"
```
Raw data
{
"_id": null,
"home_page": null,
"name": "autogluon.assistant",
"maintainer": null,
"docs_url": null,
"requires_python": "<3.12,>=3.8",
"maintainer_email": null,
"keywords": null,
"author": "AutoGluon Community",
"author_email": null,
"download_url": "https://files.pythonhosted.org/packages/29/48/c1aab5f286f116ee218b7aeabbb32f98f22cf3b5cb2adaed01670d92e472/autogluon_assistant-0.1.0.tar.gz",
"platform": null,
"description": "<div align=\"left\">\n <img src=\"https://user-images.githubusercontent.com/16392542/77208906-224aa500-6aba-11ea-96bd-e81806074030.png\" width=\"350\">\n</div>\n# AutoGluon Assistant\n\n[](https://pypi.org/project/autogluon.assistant/)\n[](./LICENSE)\n[](https://github.com/autogluon/autogluon-assistant/actions/workflows/continuous_integration.yml)\n\n\nAutoGluon Assistant (AG-A) provides users a simple interface where they can input their data, describe their problem, and receive a highly accurate and competitive ML solution \u2014 without writing any code. By leveraging the state-of-the-art AutoML capabilities of [AutoGluon](https://github.com/autogluon/autogluon) and integrating them with a Large Language Model (LLM), AG-A automates the entire data science pipeline. AG-A takes [AutoGluon](https://github.com/autogluon/autogluon)'s automation from three lines of code to zero, enabling users to solve new supervised learning tabular problems using only natural language descriptions.\n\n## \ud83d\udcbe Installation\n\nAutoGluon Assistant is supported on Python 3.8 - 3.11 and is available on Linux, MacOS, and Windows.\n\nYou can install with:\n\n```bash\npip install autogluon.assistant\n```\n\nYou can also install from source:\n\n```bash\ngit clone https://github.com/autogluon/autogluon-assistant.git\ncd autogluon-assistant && pip install -e \".\"\n```\n\n\n#### Beta Features\n\nAG-A now supports automatic feature generation as part of its beta features. To enable these features, please install the beta version dependencies using the following command:\n\n```bash\npip install -r requirements.txt\n```\n\n### API Keys\n\n#### Configuring LLMs\nAG-A supports using both AWS Bedrock and OpenAI as LLM model providers. You will need to set up API keys for the respective provider you choose. By default, AG-A uses AWS Bedrock for its language models.\n\n#### AWS Bedrock Setup\nAG-A integrates with AWS Bedrock by default. To use AWS Bedrock, you will need to configure your AWS credentials and region settings:\n\n```bash\nexport AWS_DEFAULT_REGION=\"<your-region>\"\nexport AWS_ACCESS_KEY_ID=\"<your-access-key>\"\nexport AWS_SECRET_ACCESS_KEY=\"<your-secret-key>\"\n```\n\nEnsure you have an active AWS account and appropriate permissions set up for using Bedrock models. You can manage your AWS credentials through the AWS Management Console. See [Bedrock supported AWS regions](https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-regions.html)\n\n\n#### OpenAI Setup\nTo use OpenAI, you'll need to set your OpenAI API key as an environment variable:\n\n```bash\nexport OPENAI_API_KEY=\"sk-...\"\n```\n\nYou can sign up for an OpenAI account [here](https://platform.openai.com/) and manage your API keys [here](https://platform.openai.com/account/api-keys).\n\nImportant: Free-tier OpenAI accounts may be subject to rate limits, which could affect AG-A's performance. We recommend using a paid OpenAI API key for seamless functionality.\n\n\n## Usage\n\nWe support two ways of using AutoGluon Assistant: WebUI and CLI.\n\n### Web UI\nAutoGluon Assistant Web UI allows users to leverage the capabilities of AG-A through an intuitive web interface.\n\nThe web UI enables users to upload datasets, configure AG-A runs with customized settings, preview data, monitor execution progress, view and download results, and supports secure, isolated sessions for concurrent users.\n\n#### To run the AG-A Web UI:\n\n```bash\naga ui\n\n# OR\n\n# Launch Web-UI on specific port e.g. 8888\naga ui --port 8888\n```\n\nAG-A Web UI should now be accessible in your web browser at `http://localhost:8501` or the specified port.\n\n\n### CLI\n\nBefore launching AG-A CLI, prepare your data files in the following structure:\n\n```\n\u2514\u2500\u2500 data # Data files directory\n \u251c\u2500\u2500 train.[ext] # Training dataset (required)\n \u251c\u2500\u2500 test.[ext] # Test dataset (required)\n \u2514\u2500\u2500 description.txt # Dataset and task description (recommended)\n```\n\nNote:\n- The training and test files can be in any tabular data format (e.g., csv, parquet, xlsx)\n- While there are no strict naming requirements, we recommend using clear, descriptive filenames\n- The description file is optional but recommended for better model selection and optimization. It can include:\n - Dataset description\n - Problem context\n - Evaluation metrics\n - Any other relevant information\n\nNow you can launch the AutoGluon Assistant run using the following command:\n\n```bash\naga run [NAME_OF_DATA_DIR] --presets [PRESET_QUALITY]\n# e.g. aga run ./toy_data --presets best_quality\n```\n\nWe support three presets, including `medium_quality`, `high_quality` and `best_quality`. We use `best_quality` as a default setting.\n\nAfter the run is complete, model predictions on test dataset are saved into the `aga-output-<timestamp>.csv` file. It will be formatted according to optional `sample_submission.csv` file if provided.\n\n#### Overriding Configs\n\nYou can override specific settings in the YAML configuration defined in the [config folder](https://github.com/autogluon/autogluon-assistant/tree/main/src/autogluon/assistant/configs) using\nthe `config_overrides` parameter with format `\"key1=value1, key2.nested=value2\"` from the command line.\n\n\nHere are some example commands on using configuration overrides:\n\n```bash\naga run toy_data --config_overrides \"feature_transformers.enabled_models=None, time_limit=3600\"\n\n# OR\n\naga run toy_data --config_overrides \"feature_transformers.enabled_models=None\" --config_overrides \"time_limit=3600\"\n```\n",
"bugtrack_url": null,
"license": null,
"summary": "ML Assistant for Competitive Machine Learning",
"version": "0.1.0",
"project_urls": {
"Homepage": "https://github.com/autogluon/autogluon-assistant"
},
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "d95a138cc243db0fa4a7ffda3cb8357a3630148c16f362ca2dd19bffa011786b",
"md5": "67c628aff8f76e815e10a8221022fa30",
"sha256": "4f37b44b8dbd4cc40c09a329b24d2a8ea5acb97794deaed0bdfefca5142fa64b"
},
"downloads": -1,
"filename": "autogluon.assistant-0.1.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "67c628aff8f76e815e10a8221022fa30",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<3.12,>=3.8",
"size": 1616104,
"upload_time": "2024-11-19T23:27:26",
"upload_time_iso_8601": "2024-11-19T23:27:26.831016Z",
"url": "https://files.pythonhosted.org/packages/d9/5a/138cc243db0fa4a7ffda3cb8357a3630148c16f362ca2dd19bffa011786b/autogluon.assistant-0.1.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "2948c1aab5f286f116ee218b7aeabbb32f98f22cf3b5cb2adaed01670d92e472",
"md5": "b0b357ba61d28a3fc2af2d12e142b466",
"sha256": "9f13843ccd78f02212b4a51470a2cbf85f6efd708b3907a33f8e9d6126e1103d"
},
"downloads": -1,
"filename": "autogluon_assistant-0.1.0.tar.gz",
"has_sig": false,
"md5_digest": "b0b357ba61d28a3fc2af2d12e142b466",
"packagetype": "sdist",
"python_version": "source",
"requires_python": "<3.12,>=3.8",
"size": 1608352,
"upload_time": "2024-11-19T23:27:29",
"upload_time_iso_8601": "2024-11-19T23:27:29.326231Z",
"url": "https://files.pythonhosted.org/packages/29/48/c1aab5f286f116ee218b7aeabbb32f98f22cf3b5cb2adaed01670d92e472/autogluon_assistant-0.1.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-11-19 23:27:29",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "autogluon",
"github_project": "autogluon-assistant",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"requirements": [],
"lcname": "autogluon.assistant"
}
