Metadata-Version: 2.1
Name: dria_agent
Version: 0.1.54
Summary: speedy, performant and small tool calling agents
License: MIT
Author: andthattoo
Author-email: omer@firstbatch.xyz
Requires-Python: >=3.12,<4.0
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.12
Provides-Extra: huggingface
Provides-Extra: mlx
Provides-Extra: tools
Requires-Dist: docker (>=7.1.0,<8.0.0) ; extra == "tools" or extra == "tools"
Requires-Dist: duckduckgo-search (>=7.3.2,<8.0.0) ; extra == "tools"
Requires-Dist: google-api-python-client (>=2.160.0,<3.0.0) ; extra == "tools"
Requires-Dist: google-auth-oauthlib (>=1.2.1,<2.0.0) ; extra == "tools"
Requires-Dist: markdownify (>=0.14.1,<0.15.0) ; extra == "tools"
Requires-Dist: mlx (>=0.22.1,<0.23.0) ; extra == "mlx"
Requires-Dist: mlx-lm (>=0.21.4,<0.22.0) ; extra == "mlx"
Requires-Dist: numpy (>=2.2.2,<3.0.0)
Requires-Dist: ollama (>=0.4.7,<0.5.0)
Requires-Dist: openai (>=1.61.1,<2.0.0)
Requires-Dist: python-telegram (>=0.19.0,<0.20.0) ; extra == "tools"
Requires-Dist: requests (>=2.32.3,<3.0.0)
Requires-Dist: rich (>=13.9.4,<14.0.0)
Requires-Dist: scikit-learn (>=1.6.1,<2.0.0) ; extra == "tools"
Requires-Dist: sentence-transformers (>=3.4.1,<4.0.0) ; extra == "huggingface" or extra == "mlx"
Requires-Dist: slack-sdk (>=3.34.0,<4.0.0) ; extra == "tools"
Requires-Dist: smolagents (>=1.8.1,<2.0.0) ; extra == "tools"
Requires-Dist: transformers (>=4.48.3,<5.0.0) ; extra == "huggingface"
Requires-Dist: urllib3 (>=2.3.0,<3.0.0)
Description-Content-Type: text/markdown

# Dria Agent

Dria Agent is a lightweight pip package to use fastest and most performant models for function calling on edge devices, `tiny-agent-α`.

Tiny-Agent-α is an extension of [Dria-Agent-a](https://huggingface.co/collections/driaforall/dria-agent-a-67a61f4b7d3d544fe5d3cd8a=), trained on top of the Qwen2.5-Coder series to be used in edge devices. 
These models are carefully fine-tuned with quantization aware training to minimize performance degradation after quantization. 
The smallest model is 0.5B with 4bit quantization (398MB on disk), and the largest model is 3B with 4bit quantization.


### Installation

To install the package run: (Best for CPU inference)
```bash
pip install dria_agent
pip install 'dria_agent[mlx]' # To use MLX as backend for macOS. 
pip install 'dria_agent[huggingface]' # For HuggingFace backend for GPU.
pip install 'dria_agent[mlx, tools]' # In order to use extra tooling select a backend
```

### Quick Start

Write your functions in pure python, decorate them with @tool to expose them to the agent.

````python
from dria_agent import tool

@tool
def check_availability(day: str, start_time: str, end_time: str) -> bool:
    # Your implementation here
    return True
````

Create an agent:

```python
from dria_agent import ToolCallingAgentFactory

agent = ToolCallingAgentFactory(
    tools=[check_availability], backend="ollama"
)
```

Use agent.run(query) to execute tasks with tools.

```python
query = "Check if I'm available on Monday from 10:00 to 11:00"
execution = agent.run(query, print_results=True)
```


## Models

| Model                  | Description                                | HF Download Link                                                                                                         | Ollama Tag                         | Size   |
|------------------------|--------------------------------------------|--------------------------------------------------------------------------------------------------------------------------|-------------------------------------|--------|
| Tiny-Agent-a-3B (8bit) | High performance and reasoning             | [Download](https://huggingface.co/driaforall/Tiny-Agent-a-3B/resolve/main/dria-agent-a-3b.Q8_0.gguf?download=true)       | driaforall/tiny-agent-a:3B-q8_0  | 3.3 GB |
| Tiny-Agent-a-3B (4bit) | Tradeoff 3B quality for memory             | [Download](https://huggingface.co/driaforall/Tiny-Agent-a-3B/resolve/main/dria-agent-a-3b.Q4_K_M.gguf?download=true)     | driaforall/tiny-agent-a:3B-q4_K_M | 1.9 GB |
| Tiny-Agent-a-1.5B (8bit) | Balanced performance and speed             | [Download](https://huggingface.co/driaforall/Tiny-Agent-a-1.5B/resolve/main/dria-agent-a-1.5b.Q8_0.gguf?download=true)   | driaforall/tiny-agent-a:1.5B-q8_0 | 1.6 GB |
| Tiny-Agent-a-1.5B (4bit) | Faster CPU inference, performance tradeoff | [Download](https://huggingface.co/driaforall/Tiny-Agent-a-1.5B/resolve/main/dria-agent-a-1.5b.Q8_0.gguf?download=true)   | driaforall/tiny-agent-a:1.5B-q4_K_M | 986 MB |
| Tiny-Agent-a-0.5B (8bit) | Ultra-light                                | [Download](https://huggingface.co/driaforall/Tiny-Agent-a-1.5B/resolve/main/dria-agent-a-1.5b.Q4_K_M.gguf?download=true) | driaforall/tiny-agent-a:0.5B-q8_0 | 531 MB |

a fast and powerful tool calling model designed to run on edge devices.

3B 8bit high performance and quality [download](https://huggingface.co/driaforall/Tiny-Agent-a-3B/resolve/main/dria-agent-a-3b.Q8_0.gguf?download=true)

3B 4bit balanced performance and speed [download](https://huggingface.co/driaforall/Tiny-Agent-a-3B/resolve/main/dria-agent-a-3b.Q4_K_M.gguf?download=true)

1.5B 8bit fast CPU inference [download](https://huggingface.co/driaforall/Tiny-Agent-a-1.5B/resolve/main/dria-agent-a-1.5b.Q8_0.gguf?download=true)

1.5B 4bit faster CPU inference [download](https://huggingface.co/driaforall/Tiny-Agent-a-1.5B/resolve/main/dria-agent-a-1.5b.Q8_0.gguf?download=true)

0.5B 8bit ultra-light [download](https://huggingface.co/driaforall/Tiny-Agent-a-1.5B/resolve/main/dria-agent-a-1.5b.Q4_K_M.gguf?download=true)


## Evaluation & Performance

We evaluate the model on the **Dria-Pythonic-Agent-Benchmark ([DPAB](https://github.com/firstbatchxyz/function-calling-eval)):** The benchmark we curated with a synthetic data generation +model-based validation + filtering and manual selection to evaluate LLMs on their Pythonic function calling ability, spanning multiple scenarios and tasks. See [blog](https://huggingface.co/blog/andthattoo/dpab-a) for more information.

Below are the DPAB results: 

Current benchmark results for various models **(strict)**:

| Model Name                      | Pythonic | JSON |
|---------------------------------|----------|------|
| **Closed Models**               |          |      |
| Claude 3.5 Sonnet              | 87       | 45   |
| gpt-4o-2024-11-20              | 60       | 30   |
| **Open Models**                 |          |      |
| **> 100B Parameters**           |          |      |
| DeepSeek V3 (685B)             | 63       | 33   |
| MiniMax-01                     | 62       | 40   |
| Llama-3.1-405B-Instruct        | 60       | 38   |
| **> 30B Parameters**            |          |      |
| Qwen-2.5-Coder-32b-Instruct    | 68       | 32   |
| Qwen-2.5-72b-instruct          | 65       | 39   |
| Llama-3.3-70b-Instruct         | 59       | 40   |
| QwQ-32b-Preview                | 47       | 21   |
| **< 20B Parameters**           |          |      |
| Phi-4 (14B)                    | 55       | 35   |
| Qwen2.5-Coder-7B-Instruct      | 44       | 39   |
| Qwen-2.5-7B-Instruct           | 47       | 34   |
| **Tiny-Agent-a-3B**               | **72**       | 34   |
| Qwen2.5-Coder-3B-Instruct      | 26       | 37   |
| **Tiny-Agent-a-1.5B**               | **73**       | 30   |


#### Citation

```
@misc{Dria-Agent-a,
      url={https://huggingface.co/blog/andthattoo/dria-agent-a},
      title={Dria-Agent-a},
      author={"andthattoo", "Atakan Tekparmak"}
}
```


