getllm 0.1.17

Python LLM operations service for the DevLama ecosystem

getllm

getllm is a Python package for managing LLM models with Ollama integration and generating Python code. It allows you to install, list, set the default model, update the model list, and generate code using LLM models. GetLLM is part of the PyLama ecosystem and integrates with LogLama as the primary service for centralized logging and environment management.

Features

• Code Generation: Generate Python code using LLM models
• Model Management: Install, list, and select models
• Hugging Face Integration: Search and install models from Hugging Face
• Automatic Model Installation: Automatically install models when they are not found
• Fallback Mechanisms: Use fallback models when the requested model is not available
• Environment Configuration: Configure Ollama through environment variables
• Special Model Handling: Special installation process for SpeakLeash Bielik models
• Mock Mode: Support for mock mode without requiring Ollama
• Interactive Mode: Interactive CLI for model selection and code generation
• Template System: Generate code with awareness of platform, dependencies, and more
• Code Execution: Execute generated code directly

LogLama Integration

PyLLM integrates with LogLama as the primary service in the PyLama ecosystem. This integration provides:

• Centralized Environment Management: Environment variables are loaded from the central .env file in the devlama directory
• Shared Configuration: Model configurations are shared across all PyLama components
• Dependency Management: Dependencies are validated and installed by LogLama
• Service Orchestration: Services are started in the correct order using LogLama CLI
• Centralized Logging: All PyLLM operations are logged to the central LogLama system
• Structured Logging: Logs include component context for better filtering and analysis
• Health Monitoring: LogLama monitors PyLLM service health and availability

---

General Diagram (Mermaid)

graph TD
    A[User] -->|CLI/Interactive| B[getllm/cli.py]
    B --> C[models.py]
    B --> D[interactive_cli.py]
    C --> E[LogLama Central .env]
    C --> F[Ollama API]
    D --> B
    G[LogLama] --> E

---

ASCII Diagram: CLI Command Flow

User
    |
    v
+-----------------+
|   getllm CLI     |
+-----------------+
    |
    v
+-----------------+
|   models.py     |
+-----------------+
    |
+-----------------+
| LogLama Central |
|    .env File    |
+-----------------+
    |
+-----------------+
|  Ollama API     |
+-----------------+

---

Usage

Basic Usage

# Generate code using the default model
getllm "create a function to calculate fibonacci numbers"

# Generate and run code
getllm -r "create a web server with Flask"

# Save generated code to a file
getllm -s "create a script to download files from URLs"

# Use a specific model
getllm --model codellama:7b "create a binary search tree implementation"

# Use mock mode (no Ollama required)
getllm --mock "print hello world"

# Use a model that will be automatically installed if not available
getllm --model SpeakLeash/bielik-1.5b-v3.0-instruct-gguf "print hello world"

Model Management

# List available models
getllm list

# Install a model
getllm install codellama:7b

# List installed models
getllm installed

# Set default model
getllm set-default codellama:7b

# Show default model
getllm default

# Update models list from Ollama
getllm update

Hugging Face Integration

# Search for models on Hugging Face
getllm --search bielik

# Update models list from Hugging Face
getllm --update-hf

Interactive Mode

# Start interactive mode
getllm -i

# Start interactive mode with mock implementation
getllm -i --mock

---

Installation

# Create a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install the package in development mode
pip install -e .  # This is important! Always install in development mode before starting

IMPORTANT: Always run pip install -e . before starting the project to ensure all dependencies are properly installed and the package is available in development mode.

---

Using the Makefile

PyLLM includes a Makefile to simplify common development tasks:

# Set up the project (creates a virtual environment and installs dependencies)
make setup

# Run the API server (default port 8001)
make run

# Run the API server on a custom port
make run PORT=8080

# The run-port command is also available for backward compatibility
make run-port PORT=8080

# Run tests
make test

# Format code with black
make format

# Lint code with flake8
make lint

# Clean up project (remove __pycache__, etc.)
make clean

# Show all available commands
make help

---

Key Files

• getllm/cli.py – main CLI
• getllm/interactive_cli.py – interactive shell with menu and cursor selection
• getllm/models.py – model logic, .env/env.example handling, Ollama integration
• .env/env.example – environment config and default model

---

Example Usage

Search polish moel bielik in huggingface

getllm --search bielik

from huggingface

Searching for models matching 'bielik' on Hugging Face...
Searching for models matching 'bielik' on Hugging Face...
? Select a model to install: (Use arrow keys)
 » speakleash/Bielik-11B-v2.3-Instruct-FP8            Unknown    Downloads: 26,103 |
   speakleash/Bielik-11B-v2.3-Instruct-GGUF           Unknown    Downloads: 2,203 |
   speakleash/Bielik-4.5B-v3.0-Instruct-GGUF          Unknown    Downloads: 967 |
   speakleash/Bielik-7B-Instruct-v0.1-GGUF            Unknown    Downloads: 712 |
   speakleash/Bielik-1.5B-v3.0-Instruct-GGUF          Unknown    Downloads: 423 |
   bartowski/Bielik-11B-v2.2-Instruct-GGUF            Unknown    Downloads: 382 |
   gaianet/Bielik-4.5B-v3.0-Instruct-GGUF             Unknown    Downloads: 338 |
   second-state/Bielik-1.5B-v3.0-Instruct-GGUF        Unknown    Downloads: 314 |
   second-state/Bielik-4.5B-v3.0-Instruct-GGUF        Unknown    Downloads: 306 |
   DevQuasar/speakleash.Bielik-4.5B-v3.0-Instruct-GGUF Unknown    Downloads: 219 |
   DevQuasar/speakleash.Bielik-1.5B-v3.0-Instruct-GGUF Unknown    Downloads: 219 |
   gaianet/Bielik-11B-v2.3-Instruct-GGUF              Unknown    Downloads: 173 |
   tensorblock/Bielik-11B-v2.2-Instruct-GGUF          Unknown    Downloads: 168 |
   speakleash/Bielik-11B-v2.2-Instruct-GGUF           Unknown    Downloads: 162 |
   mradermacher/Bielik-11B-v2-i1-GGUF                 Unknown    Downloads: 147 |
   gaianet/Bielik-1.5B-v3.0-Instruct-GGUF             Unknown    Downloads: 145 |
   QuantFactory/Bielik-7B-v0.1-GGUF                   Unknown    Downloads: 135 |
   second-state/Bielik-11B-v2.3-Instruct-GGUF         Unknown    Downloads: 125 |
   RichardErkhov/speakleash_-_Bielik-11B-v2.1-Instruct-gguf Unknown    Downloads: 113 |
   mradermacher/Bielik-7B-v0.1-GGUF                   Unknown    Downloads: 94 |
   Cancel

on local environment

Searching for models matching 'bielik' on Hugging Face...
Searching for models matching 'bielik' on Hugging Face...
? Select a model to install: speakleash/Bielik-1.5B-v3.0-Instruct-GGUF          Unknown    Downloads: 423 | 
? Do you want to install this model now? Yes

Detected SpeakLeash Bielik model: speakleash/Bielik-1.5B-v3.0-Instruct-GGUF
Starting special installation process...

Found existing Bielik model installation: bielik-custom-1747866289:latest
Using existing model instead of downloading again.
Increased API timeout to 120 seconds for Bielik model.
Updated .env file with model settings: ~/getllm/.env

List available models

getllm list

Install a model

getllm install deepseek-coder:6.7b

Set default model

getllm set-default deepseek-coder:6.7b

Show default model

getllm default

Update model list from Ollama

getllm update

Run interactive mode (menu, cursor selection)

getllm -i

---

set_default_model function flow (Mermaid)

flowchart TD
    S[Start] --> C{Does .env exist?}
    C -- Yes --> R[Update OLLAMA_MODEL in .env]
    C -- No --> K[Copy env.example to .env]
    K --> R
    R --> E[End]

---

Interactive mode - menu (ASCII)

+--------------------------------+
|  getllm - interactive mode       |
+--------------------------------+
| > List available models         |
|   Show default model           |
|   List installed models        |
|   Install model                |
|   Set default model            |
|   Update model list            |
|   Test default model           |
|   Exit                         |
+--------------------------------+
  (navigation: arrow keys + Enter)

---

Installation

pip install getllm

Usage

Basic Model Management

from getllm import get_models, get_default_model, set_default_model, install_model

# Get available models
models = get_models()
for model in models:
    print(f"{model['name']} - {model.get('desc', '')}")

# Get the current default model
default_model = get_default_model()
print(f"Current default model: {default_model}")

# Set a new default model
set_default_model("codellama:7b")

# Install a model
install_model("deepseek-coder:6.7b")

Direct Ollama Integration

from getllm import OllamaIntegration, get_ollama_integration, start_ollama_server

# Start the Ollama server if it's not already running
ollama = start_ollama_server()

# Or create an OllamaIntegration instance with a specific model
ollama = get_ollama_integration(model="codellama:7b")

# Check if the model is available
if ollama.check_model_availability():
    print(f"Model {ollama.model} is available")
else:
    print(f"Model {ollama.model} is not available")

    # Install the model
    if ollama.install_model(ollama.model):
        print(f"Successfully installed {ollama.model}")

# List installed models
installed_models = ollama.list_installed_models()
for model in installed_models:
    print(f"Installed model: {model['name']}")

Environment Variables

The package uses the following environment variables for Ollama integration:

• OLLAMA_PATH: Path to the Ollama executable (default: 'ollama')
• OLLAMA_MODEL: Default model to use (default: 'codellama:7b')
• OLLAMA_FALLBACK_MODELS: Comma-separated list of fallback models (default: 'codellama:7b,phi3:latest,tinyllama:latest')
• OLLAMA_AUTO_SELECT_MODEL: Whether to automatically select an available model if the requested model is not found (default: 'true')
• OLLAMA_AUTO_INSTALL_MODEL: Whether to automatically install a model when it's not found (default: 'true')
• OLLAMA_TIMEOUT: API timeout in seconds (default: '30')

These variables can be set in a .env file in the project root directory or in the system environment.

License

This project is licensed under the Apache 2.0 License (see LICENSE file).