# AKALI: Aggressive Knowledge Augmenter and Language Interface
<p align="center">
<img src="assets/logo.jpg" alt="AKALI Logo" width="512"/>
</p>
AKALI is a powerful library for language model augmentation and interfaces, designed to enhance the capabilities of AI models through strategic data augmentation and efficient task management.
## Features
- **Language Interface**: Easily integrate and interact with various language models.
- **Knowledge Augmentation**: Improve model performance through intelligent data augmentation strategies.
- **Task Management**: Flexible system for handling different NLP tasks.
- **CLI Tool**: Command-line interface for easy usage and integration.
- **Customizable**: Extend functionality with custom tasks and augmentation strategies.
## Installation
```bash
pip install akali
```
## Quick Start
Using the CLI tool
Run a language interface service directly
```bash
akali run --model alierenak/gemma-7b-akali --task EntitySentimentReasoner --host 0.0.0.0 --port 8000
```
Make a prediction
```bash
akali predict --model alierenak/gemma-7b-akali --task EntitySentimentReasoner --user-message "Turkcell hiç güzel çeken bir hat değil o yüzden Vodofone'u tercih ediyorum hem de daha ucuz"
```
Using as a Python Library
```python
from akali import LanguageInterface
li = LanguageInterface.load_model("alierenak/gemma-7b-akali")
# Set the task
li.set_task("EntitySentimentReasoner")
# Make a prediction
result = li.predict(system_text=None, user_message="Turkcell hiç güzel çeken bir hat değil o yüzden Vodofone'u tercih ediyorum hem de daha ucuz")
print(result)
```
# Advanced Usage
## Custom Task Creation
```python
from akali import LanguageInterface
from pydantic import BaseModel
from typing import List, Dict, Any
class CustomOutput(BaseModel):
entity_list: List[str]
results: List[Dict[str, str]]
def custom_processor(result: str) -> Dict[str, Any]:
# Your custom processing logic here
pass
li = LanguageInterface.load_model("your_model_id")
li.set_custom_task("CustomTask", custom_processor, CustomOutput)
```
## Knowledge Augmentation
```python
from akali import LanguageInterface, AggressiveKnowledgeAugmenter, AugmenterConfig
mentee = LanguageInterface.load_model("mentee_model_path")
mentor = LanguageInterface.load_model("mentor_model_path")
config = AugmenterConfig(
# Configure your augmentation settings
)
augmenter = AggressiveKnowledgeAugmenter(mentee, mentor, "EntitySentimentReasoner", config)
# Evaluate current performance
augmenter.evaluate(test_data)
# Propose augmentation strategies
strategies = augmenter.propose()
# Augment the data
augmented_data = augmenter.augment(train_data)
```
## Project Structure
```
akali/
│
├── src/
│ └── akali/
│ ├── __init__.py
│ ├── aka.py
│ ├── li.py
│ ├── utils.py
│ ├── task_manager.py
│ ├── models/
│ ├── processors/
│ └── ...
│
├── examples/
├── tests/
├── .gitignore
├── pyproject.toml
├── README.md
├── requirements.txt
└── setup.py
```
## Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
## License
This project is proprietary and confidential. Unauthorized copying, transferring or reproduction of this project, via any medium, is strictly prohibited.
## Contact
For licensing inquiries, support or anything, here is my contact mail: akali@sabanciuniv.edu.
Raw data
{
"_id": null,
"home_page": "https://github.com/alierenak/akali",
"name": "akali",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": null,
"keywords": "NLP, language models, data augmentation, AI",
"author": "Ali Eren Ak",
"author_email": "akali@sabanciuniv.edu",
"download_url": "https://files.pythonhosted.org/packages/83/6d/d2b316fecf620d58247474c7e0a782387e096d9a42d5a98db5e33b4abb80/akali-0.3.1b0.tar.gz",
"platform": null,
"description": "# AKALI: Aggressive Knowledge Augmenter and Language Interface\n\n<p align=\"center\">\n <img src=\"assets/logo.jpg\" alt=\"AKALI Logo\" width=\"512\"/>\n</p>\n\nAKALI is a powerful library for language model augmentation and interfaces, designed to enhance the capabilities of AI models through strategic data augmentation and efficient task management.\n\n## Features\n\n- **Language Interface**: Easily integrate and interact with various language models.\n- **Knowledge Augmentation**: Improve model performance through intelligent data augmentation strategies.\n- **Task Management**: Flexible system for handling different NLP tasks.\n- **CLI Tool**: Command-line interface for easy usage and integration.\n- **Customizable**: Extend functionality with custom tasks and augmentation strategies.\n\n## Installation\n\n```bash\npip install akali\n```\n\n## Quick Start\nUsing the CLI tool\nRun a language interface service directly\n```bash\nakali run --model alierenak/gemma-7b-akali --task EntitySentimentReasoner --host 0.0.0.0 --port 8000\n```\n\nMake a prediction\n```bash\nakali predict --model alierenak/gemma-7b-akali --task EntitySentimentReasoner --user-message \"Turkcell hi\u00e7 g\u00fczel \u00e7eken bir hat de\u011fil o y\u00fczden Vodofone'u tercih ediyorum hem de daha ucuz\"\n```\n\nUsing as a Python Library\n```python\nfrom akali import LanguageInterface\nli = LanguageInterface.load_model(\"alierenak/gemma-7b-akali\")\n\n# Set the task\nli.set_task(\"EntitySentimentReasoner\")\n\n# Make a prediction\nresult = li.predict(system_text=None, user_message=\"Turkcell hi\u00e7 g\u00fczel \u00e7eken bir hat de\u011fil o y\u00fczden Vodofone'u tercih ediyorum hem de daha ucuz\")\nprint(result)\n```\n# Advanced Usage\n\n## Custom Task Creation\n```python\nfrom akali import LanguageInterface\nfrom pydantic import BaseModel\nfrom typing import List, Dict, Any\n\nclass CustomOutput(BaseModel):\n entity_list: List[str]\n results: List[Dict[str, str]]\n\ndef custom_processor(result: str) -> Dict[str, Any]:\n # Your custom processing logic here\n pass\n\nli = LanguageInterface.load_model(\"your_model_id\")\nli.set_custom_task(\"CustomTask\", custom_processor, CustomOutput)\n```\n## Knowledge Augmentation\n\n```python\nfrom akali import LanguageInterface, AggressiveKnowledgeAugmenter, AugmenterConfig\n\nmentee = LanguageInterface.load_model(\"mentee_model_path\")\nmentor = LanguageInterface.load_model(\"mentor_model_path\")\n\nconfig = AugmenterConfig(\n # Configure your augmentation settings\n)\n\naugmenter = AggressiveKnowledgeAugmenter(mentee, mentor, \"EntitySentimentReasoner\", config)\n\n# Evaluate current performance\naugmenter.evaluate(test_data)\n\n# Propose augmentation strategies\nstrategies = augmenter.propose()\n\n# Augment the data\naugmented_data = augmenter.augment(train_data)\n```\n\n## Project Structure\n\n```\nakali/\n\u2502\n\u251c\u2500\u2500 src/\n\u2502 \u2514\u2500\u2500 akali/\n\u2502 \u251c\u2500\u2500 __init__.py\n\u2502 \u251c\u2500\u2500 aka.py\n\u2502 \u251c\u2500\u2500 li.py\n\u2502 \u251c\u2500\u2500 utils.py\n\u2502 \u251c\u2500\u2500 task_manager.py\n\u2502 \u251c\u2500\u2500 models/\n\u2502 \u251c\u2500\u2500 processors/\n\u2502 \u2514\u2500\u2500 ...\n\u2502\n\u251c\u2500\u2500 examples/\n\u251c\u2500\u2500 tests/\n\u251c\u2500\u2500 .gitignore\n\u251c\u2500\u2500 pyproject.toml\n\u251c\u2500\u2500 README.md\n\u251c\u2500\u2500 requirements.txt\n\u2514\u2500\u2500 setup.py\n```\n## Contributing\n\nContributions are welcome! Please feel free to submit a Pull Request.\n\n## License\n\nThis project is proprietary and confidential. Unauthorized copying, transferring or reproduction of this project, via any medium, is strictly prohibited.\n\n## Contact\n\nFor licensing inquiries, support or anything, here is my contact mail: akali@sabanciuniv.edu.\n\n\n\n\n\n",
"bugtrack_url": null,
"license": "Proprietary",
"summary": "AKALI library for language model augmentation and interfaces",
"version": "0.3.1b0",
"project_urls": {
"Homepage": "https://github.com/alierenak/akali"
},
"split_keywords": [
"nlp",
" language models",
" data augmentation",
" ai"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "a77258708d1a5f2e869e3e641082305e3635b8358835caa2863ed2cd46d63c01",
"md5": "65344298147ee9027bf8b386cede931f",
"sha256": "69e5b125d51b219910141f805f85d169e8156cde946173d5b40ebd0e509c6305"
},
"downloads": -1,
"filename": "akali-0.3.1b0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "65344298147ee9027bf8b386cede931f",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 17877,
"upload_time": "2024-08-09T08:58:38",
"upload_time_iso_8601": "2024-08-09T08:58:38.578203Z",
"url": "https://files.pythonhosted.org/packages/a7/72/58708d1a5f2e869e3e641082305e3635b8358835caa2863ed2cd46d63c01/akali-0.3.1b0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "836dd2b316fecf620d58247474c7e0a782387e096d9a42d5a98db5e33b4abb80",
"md5": "595c4c0d9dc2300d3fe009c574aba522",
"sha256": "c961fdb501aba8bf95659b446c1a483c8145d8ed56d38a5a64dfeb99699dd36c"
},
"downloads": -1,
"filename": "akali-0.3.1b0.tar.gz",
"has_sig": false,
"md5_digest": "595c4c0d9dc2300d3fe009c574aba522",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 14493,
"upload_time": "2024-08-09T08:58:40",
"upload_time_iso_8601": "2024-08-09T08:58:40.069824Z",
"url": "https://files.pythonhosted.org/packages/83/6d/d2b316fecf620d58247474c7e0a782387e096d9a42d5a98db5e33b4abb80/akali-0.3.1b0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-08-09 08:58:40",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "alierenak",
"github_project": "akali",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"requirements": [],
"lcname": "akali"
}
JSON
