feat: Improve dependency management and installation documentation

README.md

@@ -60,14 +60,36 @@ We provide three templates, which were used in our paper, covering the following
 
 ## Requirements
 
-This code is designed to run on Linux with NVIDIA GPUs using CUDA and PyTorch. Support for other GPU architectures may be possible by following the [PyTorch guidelines](https://pytorch.org/get-started/locally/). The current templates would likely take an infeasible amount of time on CPU-only machines. Running on other operating systems may require significant adjustments.
+Please see [Hardware Requirements](docs/hardware_requirements.md) for detailed system specifications.
+
+### Prerequisites
+- Python 3.8-3.11
+- pip (latest version)
+- Virtual environment (recommended)
+- Linux operating system
+- NVIDIA GPU (optional, recommended for local models)
 
 ### Installation
 
+We support both conda and venv for environment management. Choose the method that best suits your needs:
+
+#### Option 1: Using conda (Recommended for GPU Support)
 ```bash
 conda create -n ai_scientist python=3.11
 conda activate ai_scientist
-# Install pdflatex
+```
+
+#### Option 2: Using venv
+```bash
+python -m venv venv
+source venv/bin/activate  # Linux/Mac
+# or
+.\venv\Scripts\activate  # Windows
+```
+
+#### Common Steps
+```bash
+# Install pdflatex (required for paper generation)
 sudo apt-get install texlive-full
 
 # Install PyPI requirements
@@ -76,6 +98,23 @@ pip install -r requirements.txt
 
 **Note:** Installing `texlive-full` can take a long time. You may need to [hold Enter](https://askubuntu.com/questions/956006/pregenerating-context-markiv-format-this-may-take-some-time-takes-forever) during the installation.
 
+### Model Configuration
+
+Choose from the following model options:
+
+1. Cloud API Models (Recommended for most users)
+   - OpenAI GPT-4/3.5
+   - Google Gemini Pro
+   - Anthropic Claude
+   - DeepSeek Coder V2
+
+2. Local Models (via Ollama)
+   - LLaMA 3.2/3.3
+   - Mistral
+   - Code LLaMA
+
+See [Model Configuration](docs/model_configuration.md) for detailed setup instructions.
+
 ### Supported Models and API Keys
 
 We support a wide variety of models, including open-weight and API-only models. In general, we recommend using only frontier models above the capability of the original GPT-4. To see a full list of supported models, see [here](https://github.com/SakanaAI/AI-Scientist/blob/main/ai_scientist/llm.py).

docs/hardware_requirements.md

@@ -0,0 +1,57 @@
+# Hardware Requirements
+
+This document outlines the hardware requirements for running AI-Scientist effectively.
+
+## Minimum Requirements
+
+These are the minimum specifications needed to run AI-Scientist with cloud-based models:
+
+- **RAM**: 8GB
+  - Required for basic operation and model API usage
+  - Sufficient for running with cloud-based models
+
+- **Storage**: 10GB
+  - 2GB for base installation
+  - 5GB for dependencies
+  - 3GB for workspace and generated content
+
+- **CPU**: 4 cores
+  - Recommended minimum for concurrent operations
+  - Suitable for basic research tasks
+
+- **Python**: 3.8-3.11
+  - Required for compatibility with dependencies
+  - Latest patch version within range recommended
+
+## Recommended Requirements
+
+These specifications are recommended for optimal performance, especially when using local models:
+
+- **RAM**: 16GB
+  - Recommended for running local models
+  - Provides better performance for concurrent operations
+  - Required for larger research projects
+
+- **Storage**: 20GB
+  - 2GB for base installation
+  - 8GB for dependencies
+  - 5GB for local model files
+  - 5GB for workspace and generated content
+
+- **GPU**: NVIDIA with 8GB VRAM (for local models)
+  - Required only if running local models
+  - NVIDIA GPU recommended for compatibility
+  - Minimum 8GB VRAM for standard model variants
+  - Required for optimal performance with local LLMs
+
+- **CUDA**: 11.8+
+  - Required only if using GPU
+  - Compatible with PyTorch and most ML frameworks
+  - Latest version recommended within compatibility range
+
+## Notes
+
+- GPU requirements are optional and only necessary for running local models
+- Cloud-based usage can run effectively on minimum specifications
+- Storage requirements may vary based on the number of local models installed
+- For development work, recommended specifications are strongly advised

docs/model_configuration.md

@@ -0,0 +1,153 @@
+# Model Configuration Guide
+
+This guide provides detailed instructions for setting up and configuring different model options in AI-Scientist.
+
+## Cloud API Models
+
+### 1. OpenAI Models
+- **Required Resources**
+  - RAM: 4GB minimum
+  - Storage: 2GB for dependencies
+  - Internet connection required
+- **Available Models**
+  - gpt-4-turbo (recommended)
+  - gpt-3.5-turbo
+- **Setup Instructions**
+  1. Obtain API key from [OpenAI Platform](https://platform.openai.com)
+  2. Set environment variable:
+     ```bash
+     export OPENAI_API_KEY="your-key-here"
+     ```
+
+### 2. Google Gemini Pro
+- **Required Resources**
+  - RAM: 4GB minimum
+  - Storage: 2GB for dependencies
+  - Internet connection required
+- **Available Models**
+  - gemini-pro
+- **Setup Instructions**
+  1. Set up Google Cloud project
+  2. Enable Vertex AI API
+  3. Set environment variables:
+     ```bash
+     export CLOUD_ML_REGION="your-region"
+     export VERTEXAI_PROJECT="your-project-id"
+     ```
+
+### 3. Anthropic Claude
+- **Required Resources**
+  - RAM: 4GB minimum
+  - Storage: 2GB for dependencies
+  - Internet connection required
+- **Available Models**
+  - claude-3-sonnet (recommended)
+  - claude-3-opus
+- **Setup Instructions**
+  1. Obtain API key from Anthropic
+  2. Set environment variable:
+     ```bash
+     export ANTHROPIC_API_KEY="your-key-here"
+     ```
+
+### 4. DeepSeek Coder
+- **Required Resources**
+  - RAM: 4GB minimum
+  - Storage: 2GB for dependencies
+  - Internet connection required
+- **Available Models**
+  - deepseek-coder-v2
+- **Setup Instructions**
+  1. Obtain API key from DeepSeek
+  2. Set environment variable:
+     ```bash
+     export DEEPSEEK_API_KEY="your-key-here"
+     ```
+
+## Local Models (via Ollama)
+
+### Prerequisites
+- Ollama installed ([Installation Guide](https://ollama.ai/download))
+- System requirements vary by model size
+
+### 1. LLaMA Models
+- **Required Resources**
+  - RAM: 8GB minimum (16GB recommended)
+  - Storage: 5-15GB depending on model size
+  - GPU: Optional, 8GB VRAM recommended
+- **Available Models**
+  - llama3.2:1b (minimum requirements)
+  - llama3.2:7b (recommended)
+  - llama3.3:7b (recommended)
+- **Setup Instructions**
+  1. Install Ollama
+  2. Pull model:
+     ```bash
+     ollama pull llama3.2:7b
+     ```
+  3. Start Ollama server:
+     ```bash
+     ollama serve
+     ```
+
+### 2. Mistral Models
+- **Required Resources**
+  - RAM: 6GB minimum (12GB recommended)
+  - Storage: 4-8GB
+  - GPU: Optional, 8GB VRAM recommended
+- **Available Models**
+  - mistral:7b
+- **Setup Instructions**
+  1. Install Ollama
+  2. Pull model:
+     ```bash
+     ollama pull mistral:7b
+     ```
+  3. Start Ollama server:
+     ```bash
+     ollama serve
+     ```
+
+### 3. Code LLaMA
+- **Required Resources**
+  - RAM: 8GB minimum (16GB recommended)
+  - Storage: 5-15GB
+  - GPU: Optional, 8GB VRAM recommended
+- **Available Models**
+  - codellama:7b
+- **Setup Instructions**
+  1. Install Ollama
+  2. Pull model:
+     ```bash
+     ollama pull codellama:7b
+     ```
+  3. Start Ollama server:
+     ```bash
+     ollama serve
+     ```
+
+## Troubleshooting
+
+### Common Issues
+
+1. **Insufficient Memory**
+   - Try using smaller model variants
+   - Close unnecessary applications
+   - For local models, use CPU-only mode if GPU memory is insufficient
+
+2. **API Rate Limits**
+   - Implement exponential backoff (built into our client)
+   - Consider using multiple API keys
+   - Monitor usage through provider dashboards
+
+3. **Local Model Performance**
+   - Use GPU acceleration when available
+   - Consider cloud API models for better performance
+   - Try smaller model variants if experiencing issues
+
+### Getting Help
+
+If you encounter issues not covered here:
+1. Check the [GitHub Issues](https://github.com/SakanaAI/AI-Scientist/issues)
+2. Review error messages in logs
+3. Create a new issue with detailed system information

launch_scientist.py

@@ -19,8 +19,36 @@
 from ai_scientist.perform_review import perform_review, load_paper, perform_improvement
 from ai_scientist.perform_writeup import perform_writeup, generate_latex
 
+# Constants
 NUM_REFLECTIONS = 3
 
+def check_dependencies():
+    """Verify system requirements and dependencies."""
+    # Check Python version
+    python_version = sys.version_info
+    if python_version.major != 3 or not (8 <= python_version.minor <= 11):
+        raise RuntimeError("Python version must be between 3.8 and 3.11 (inclusive)")
+
+    # Check RAM
+    try:
+        import psutil
+        total_ram = psutil.virtual_memory().total
+        if total_ram < 8 * 1024 * 1024 * 1024:  # 8GB
+            print("Warning: Less than 8GB RAM available")
+        print(f"Available RAM: {total_ram / (1024**3):.1f} GB")
+    except ImportError:
+        print("Warning: psutil not installed, RAM check skipped")
+
+    # Check GPU if available
+    try:
+        if torch.cuda.is_available():
+            print(f"GPU detected: {torch.cuda.get_device_name(0)}")
+            print(f"GPU Memory: {torch.cuda.get_device_properties(0).total_memory / (1024**3):.1f} GB")
+        else:
+            print("Warning: No GPU detected, using CPU only")
+    except Exception as e:
+        print(f"Warning: Error checking GPU: {e}")
+
 
 def print_time():
     print(datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
@@ -313,6 +341,9 @@ def do_idea(
 
 
 if __name__ == "__main__":
+    # Verify system requirements
+    check_dependencies()
+
     args = parse_arguments()
 
     # Check available GPUs and adjust parallel processes if necessary

pyproject.toml

@@ -0,0 +1,22 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "ai_scientist"
+version = "0.1.0"
+authors = [
+    { name="SakanaAI", email="[email protected]" },
+]
+description = "AI Scientist - Multi-agent AI authoring framework"
+readme = "README.md"
+requires-python = ">=3.8,<3.12"
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "License :: OSI Approved :: MIT License",
+    "Operating System :: OS Independent",
+]
+
+[project.urls]
+"Homepage" = "https://github.com/SakanaAI/AI-Scientist"
+"Bug Tracker" = "https://github.com/SakanaAI/AI-Scientist/issues"

requirements.txt

@@ -1,18 +1,37 @@
 # LLM APIs
-anthropic
-aider-chat
-backoff
-openai
-google-generativeai
+anthropic>=0.5.0
+backoff>=2.2.0
+google-cloud-aiplatform>=1.35.0
+litellm>=1.10.0
+openai>=1.0.0
+python-dotenv>=1.0.0
+
 # Viz
-matplotlib
-pypdf
-pymupdf4llm
+matplotlib>=3.7.0
+pypdf>=3.0.0
+pymupdf>=1.23.0
+pymupdf4llm>=0.0.17
+
 # Common Requirements
-torch
-numpy
-transformers
-datasets
-tiktoken
-wandb
-tqdm
+torch>=2.0.0
+numpy>=1.24.0
+transformers>=4.30.0
+datasets>=2.14.0
+tiktoken>=0.5.0
+wandb>=0.15.0
+tqdm>=4.65.0
+semanticscholar>=0.5.0
+aider-chat>=0.18.0
+
+# System Monitoring
+psutil>=5.9.0
+
+# Optional Dependencies
+ollama>=0.1.0
+
+# Development Dependencies
+pytest>=7.0.0
+black>=22.0.0
+isort>=5.10.0
+flake8>=4.0.0
+mypy>=0.990