llmfusion 0.2.2

Unified interface for multiple LLM providers

Here's a comprehensive README.md file for your LLM Providers library:

LLM Providers

A unified Python library for interacting with multiple Large Language Model (LLM) providers with a consistent interface.

Features

• Multi-Provider Support: OpenAI, Claude, Gemini, Groq, DeepSeek, HuggingFace (API and local)
• Unified Interface: Consistent API across all providers
• Async Support: Full async/await implementation
• Streaming: Real-time text generation
• Caching: Disk-based caching with TTL
• Rate Limiting: Built-in rate limiting
• Error Handling: Comprehensive error handling and retries
• Type Safety: Pydantic models for input validation
• Extensible: Easy to add new providers

Installation

pip install llm-providers

Supported Providers

Provider	API Support	Local Models	Streaming	Async
OpenAI	✅	❌	✅	✅
Claude	✅	❌	❌	✅
Gemini	✅	❌	❌	✅
Groq	✅	❌	✅	✅
DeepSeek	✅	❌	✅	✅
HuggingFace	❌	❌	❌	❌

Quick Start

Basic Usage

from llmfusion import get_provider, LLMConfig, LLMInput

# Configure OpenAI
openai_config = LLMConfig(
    api_key="your-openai-key",
    model_name="gpt-4"
)

# Create client
openai = get_provider("openai", openai_config)

# Create input
input = LLMInput(
    prompt="Explain quantum computing",
    system_prompt="You are a physics expert",
    temperature=0.7,
    max_tokens=500
)

# Generate response
response = openai.generate(input)
print(response)

Async Usage

import asyncio
from llmfusion import get_provider, LLMConfig, LLMInput


async def main():
    config = LLMConfig(
        api_key="your-api-key",
        model_name="claude-3"
    )
    claude = get_provider("claude", config)

    input = LLMInput(prompt="Write a poem about AI")
    response = await claude.agenerate(input)
    print(response)


asyncio.run(main())

Streaming

from llmfusion import get_provider, LLMConfig, LLMInput

config = LLMConfig(
    api_key="your-api-key",
    model_name="gpt-4"
)
gpt = get_provider("openai", config)

input = LLMInput(prompt="Explain blockchain technology")
for chunk in gpt.stream(input):
    print(chunk, end="", flush=True)

Configuration

LLMConfig Parameters

Parameter	Type	Default	Description
api_key	str	None	Provider API key
model_name	str	Required	Model name/identifier
base_url	str	Provider default	Custom API endpoint
timeout	int	30	Request timeout in seconds
max_retries	int	3	Maximum retry attempts
cache_ttl	int	3600	Cache time-to-live in seconds
rate_limit_rpm	int	1000	Requests per minute limit
device	str	"auto"	Device for local models (cpu/cuda)

Advanced Usage

Custom Cache Directory

from llmfusion import set_cache_dir

set_cache_dir("/path/to/cache")

Disable Caching

config = LLMConfig(
    api_key="your-key",
    model_name="gpt-4",
    cache_ttl=0  # Disable caching
)

Using Local HuggingFace Models

config = LLMConfig(
    model_name="mistralai/Mistral-7B-Instruct-v0.1"
)
hf = get_provider("huggingface", config)

Error Handling

The library provides comprehensive error handling with custom exceptions:

from llmfusion import (
    LLMError,
    RateLimitError,
    AuthenticationError,
    ProviderError
)

try:
    response = client.generate(input)
except RateLimitError as e:
    print("Rate limit exceeded:", str(e))
except AuthenticationError as e:
    print("Authentication failed:", str(e))
except ProviderError as e:
    print("Provider error:", str(e))
except LLMError as e:
    print("General error:", str(e))

Contributing

We welcome contributions! Please see our Contributing Guide for details.

Documentation

Full documentation is available at https://github.com/mbenhaddou/llm_providers

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgements

• Inspired by the need for a unified LLM interface
• Built with ❤️ by Mohamed Ben Haddou