ai-nk-cce 0.1.0__tar.gz

ai_nk_cce-0.1.0/PKG-INFO

@@ -0,0 +1,118 @@
+Metadata-Version: 2.3
+Name: ai-nk-cce
+Version: 0.1.0
+Summary: This repository is used to train AI agents to predict good strategies in a social learning game based on a NK landscape.
+Author: Luis Mienhardt
+Author-email: mienhardt@mpib-berlin.mpg.de
+Requires-Python: >=3.10,<4.0
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Provides-Extra: model
+Requires-Dist: dataframe-image (>=0.2.7,<0.3.0)
+Requires-Dist: evaluate (>=0.4.3,<0.5.0)
+Requires-Dist: matplotlib (>=3.9.2,<4.0.0)
+Requires-Dist: networkx (>=3.4.2,<4.0.0)
+Requires-Dist: numpy (>=1.26.0,<2.0.0)
+Requires-Dist: openai (>=1.52.0,<2.0.0)
+Requires-Dist: pandas (>=2.2.3,<3.0.0)
+Requires-Dist: pydantic (>=2.0.0,<3.0.0)
+Requires-Dist: python-dotenv (>=1.0.1,<2.0.0)
+Requires-Dist: pyyaml (>=6.0.2,<7.0.0)
+Requires-Dist: scikit-learn (>=1.5.2,<2.0.0)
+Requires-Dist: tiktoken (>=0.8.0,<0.9.0)
+Requires-Dist: tqdm (>=4.66.0,<5.0.0)
+Description-Content-Type: text/markdown
+
+# human-ai-social-learning
+
+[![Code style: black](https://img.shields.io/badge/code%20style-black-000000.svg)](https://github.com/psf/black) [![PEP8](https://img.shields.io/badge/code%20style-pep8-orange.svg)](https://www.python.org/dev/peps/pep-0008/)
+
+## Description
+
+## Installation
+
+This project uses Poetry for package management. If you haven't installed Poetry yet, please follow the instructions on the [official Poetry website](https://python-poetry.org/docs/#installation).
+
+To install the project:
+
+1. Clone the repository:
+
+   ```bash
+   git clone https://github.com/your-username/human-ai-social-learning.git
+   cd human-ai-social-learning
+   ```
+
+2. Install dependencies with Poetry:
+
+   ```bash
+   poetry install
+   ```
+
+3. Activate the virtual environment:
+
+   ```bash
+   poetry shell
+   ```
+
+4. (Optional) Set up Jupyter kernel for notebooks:
+
+   ```bash
+   poetry run python -m ipykernel install --user --name nk-cce-kernel
+   ```
+
+Now your development environment is set up and ready to use.
+
+## Contributing
+
+We welcome contributions! Please note:
+
+1. Please create a descriptive branch for each contribution (naming convention *feature_type/feature_name*)
+2. Follow the project style (PEP 8 for Python).
+3. Add tests and run all before commit. (At least one test per function or method.)
+4. Write meaningful commit messages.
+5. Keep in line with pre-commit linting.
+6. Submit a Pull Request, to include your code into main.
+
+### Pre-commit Hooks
+
+We use pre-commit hooks. Installation:
+
+```bash
+poetry add pre-commit
+pre-commit install
+pre-commit run --all-files
+```
+
+### Running Tests
+
+Tests need to be run in the virtual environment. You can use Poetry or Visual Studio Code settings to do so automatically.
+
+To run all tests using Poetry run:
+
+```bash
+poetry run pytest
+```
+
+We included Visual Studio Code settings in the repository.
+You can try to use them to run the tests within Visual Studio Code.
+
+```text
+.vscode/
+├── settings.json
+├── launch.json
+```
+
+## Documentation
+
+For an overview of the repository structure, module organization, and
+dependencies, see the [Repository Structure Guide](doc/repository_structure.md).
+
+Additional documentation is available in the `doc/` directory:
+
+- [How to compare hill climber vs AI](doc/how_to_compare_hill_climber_vs_ai.md)
+- [How to find average hill climber landscape](doc/how_to_find_average_hill_climber_landscape.md)
+- [Using MPCDF LLM inference](doc/use_mpcdf_llm_inference.md)
+

ai_nk_cce-0.1.0/README.md

@@ -0,0 +1,89 @@
+# human-ai-social-learning
+
+[![Code style: black](https://img.shields.io/badge/code%20style-black-000000.svg)](https://github.com/psf/black) [![PEP8](https://img.shields.io/badge/code%20style-pep8-orange.svg)](https://www.python.org/dev/peps/pep-0008/)
+
+## Description
+
+## Installation
+
+This project uses Poetry for package management. If you haven't installed Poetry yet, please follow the instructions on the [official Poetry website](https://python-poetry.org/docs/#installation).
+
+To install the project:
+
+1. Clone the repository:
+
+   ```bash
+   git clone https://github.com/your-username/human-ai-social-learning.git
+   cd human-ai-social-learning
+   ```
+
+2. Install dependencies with Poetry:
+
+   ```bash
+   poetry install
+   ```
+
+3. Activate the virtual environment:
+
+   ```bash
+   poetry shell
+   ```
+
+4. (Optional) Set up Jupyter kernel for notebooks:
+
+   ```bash
+   poetry run python -m ipykernel install --user --name nk-cce-kernel
+   ```
+
+Now your development environment is set up and ready to use.
+
+## Contributing
+
+We welcome contributions! Please note:
+
+1. Please create a descriptive branch for each contribution (naming convention *feature_type/feature_name*)
+2. Follow the project style (PEP 8 for Python).
+3. Add tests and run all before commit. (At least one test per function or method.)
+4. Write meaningful commit messages.
+5. Keep in line with pre-commit linting.
+6. Submit a Pull Request, to include your code into main.
+
+### Pre-commit Hooks
+
+We use pre-commit hooks. Installation:
+
+```bash
+poetry add pre-commit
+pre-commit install
+pre-commit run --all-files
+```
+
+### Running Tests
+
+Tests need to be run in the virtual environment. You can use Poetry or Visual Studio Code settings to do so automatically.
+
+To run all tests using Poetry run:
+
+```bash
+poetry run pytest
+```
+
+We included Visual Studio Code settings in the repository.
+You can try to use them to run the tests within Visual Studio Code.
+
+```text
+.vscode/
+├── settings.json
+├── launch.json
+```
+
+## Documentation
+
+For an overview of the repository structure, module organization, and
+dependencies, see the [Repository Structure Guide](doc/repository_structure.md).
+
+Additional documentation is available in the `doc/` directory:
+
+- [How to compare hill climber vs AI](doc/how_to_compare_hill_climber_vs_ai.md)
+- [How to find average hill climber landscape](doc/how_to_find_average_hill_climber_landscape.md)
+- [Using MPCDF LLM inference](doc/use_mpcdf_llm_inference.md)

ai_nk_cce-0.1.0/pyproject.toml

@@ -0,0 +1,135 @@
+[tool.poetry]
+name = "ai-nk-cce"
+version = "0.1.0"
+description = "This repository is used to train AI agents to predict good strategies in a social learning game based on a NK landscape."
+authors = [
+"Luis Mienhardt <mienhardt@mpib-berlin.mpg.de>",
+"Levin Brinkmann <brinkmann@mpib-berlin.mpg.de>",
+"Bramantyo Ibrahim Supriyatno <supriyatno@mpib-berlin.mpg.de>",
+]
+readme = "README.md"
+packages = [
+    { include = "api", from = "src" },
+    { include = "evals", from = "src" },
+    { include = "model", from = "src" },
+    { include = "nk_model", from = "src" },
+    { include = "simulation", from = "src" },
+    { include = "utils", from = "src" },
+]
+
+[tool.poetry.dependencies]
+numpy = "^1.26.0"
+pandas = "^2.2.3"
+pyyaml = "^6.0.2"
+python = "^3.10"
+matplotlib = "^3.9.2"
+networkx = "^3.4.2"
+openai = "^1.52.0"
+python-dotenv = "^1.0.1"
+scikit-learn = "^1.5.2"
+evaluate = "^0.4.3"
+tiktoken = "^0.8.0"
+dataframe-image = "^0.2.7"
+pydantic = "^2.0.0"
+tqdm = "^4.66.0"
+
+[tool.poetry.extras]
+model = [
+    "seaborn",
+    "accelerate",
+    "transformers",
+    "tokenizers",
+    "torch",
+    "datasets",
+    "wandb",
+    "trl",
+    "git-lfs",
+]
+
+[tool.poetry.group.dev.dependencies]
+black = "^24.10.0"
+pylint = "^3.3.1"
+pytest = "^8.3.3"
+pre-commit = "^4.0.1"
+pytest-mock = "^3.14.0"
+coverage = "^7.10.3"
+ipykernel = "^6.0.0"
+jupyter = "^1.1.1"
+pytest-asyncio = "^0.25.3"
+tqdm = "^4.66.0"
+
+[tool.poetry.group.model.dependencies]
+seaborn = "^0.13.2"
+accelerate = "^1.0.1"
+transformers = "^4.57.3"
+tokenizers = "^0.22.1"
+torch = "^2.1.0"
+datasets = "4.0.0"
+wandb = "^0.18.7"
+trl = "^0.18.1"
+git-lfs = "^1.6"
+
+
+[tool.black]
+line-length = 79
+target-version = ['py310']
+include = '\.pyi?$'
+extend-exclude = '''
+/(
+  # directories
+  \.eggs
+  | \.git
+  | \.hg
+  | \.mypy_cache
+  | \.tox
+  | \.venv
+  | build
+  | dist
+)/
+'''
+
+[tool.isort]
+profile = "black"
+line_length = 79
+
+[tool.flake8]
+max-line-length = 79
+extend-ignore = "E203, W503"
+max-complexity = 10
+
+[tool.pytest.ini_options]
+testpaths = ["."]
+python_files = "test_*.py"
+python_classes = "Test*"
+python_functions = "test_*"
+addopts = "-v -ra -s"
+pythonpath = [
+  "src"
+]
+
+[tool.coverage.run]
+source = ["src"]
+omit = [
+    "*/tests/*",
+    "*/test_*.py",
+    "*/__pycache__/*",
+    "*/conftest.py",
+]
+
+[tool.coverage.report]
+exclude_lines = [
+    "pragma: no cover",
+    "def __repr__",
+    "raise AssertionError",
+    "raise NotImplementedError",
+    "if __name__ == .__main__.:",
+    "if TYPE_CHECKING:",
+    "@abstractmethod",
+]
+precision = 2
+show_missing = true
+skip_covered = false
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"

ai_nk_cce-0.1.0/src/api/mpcdf_vllm.py

@@ -0,0 +1,94 @@
+import os
+import time
+from typing import Any, Optional
+
+import httpx
+from dotenv import load_dotenv
+
+from src.utils.logging import log_dec
+
+load_dotenv()
+
+BASE_URL = os.getenv("MPCDF_BASE_URL")
+API_KEY = os.getenv("MPCDF_API_KEY")
+MODEL_NAME = "center-for-humans-and-machines/assembl_nk_1"
+
+
+@log_dec
+def mpcdf_vllm_request(
+    prompt: str,
+    model: Optional[str] = None,
+    max_tokens: int = 150,
+    temperature: float = 0.7,
+    timeout: float = 30.0,
+    max_retries: int = 3,
+    initial_backoff: float = 1.0,
+    **kwargs: Any,
+) -> str:
+    """Make request to vLLM model with tokenized input.
+
+    Uses the /v1/completions endpoint which supports prompt_token_ids
+    directly, bypassing the tokenizer. This function tokenizes the prompt
+    and sends it as token IDs.
+
+    Args:
+        prompt: Text prompt string to send to the model.
+        model: Model name (defaults to MODEL_NAME).
+        max_tokens: Maximum tokens to generate.
+        temperature: Sampling temperature.
+        timeout: Request timeout in seconds.
+        max_retries: Maximum number of retry attempts (default: 3).
+        initial_backoff: Initial backoff delay in seconds (default: 1.0).
+        **kwargs: Additional parameters for completion request.
+
+    Returns:
+        Generated text content.
+    """
+    # Build request URL - ensure /v1 prefix if not present
+    base = BASE_URL.rstrip("/")
+    if not base.endswith("/v1"):
+        base = f"{base}/v1"
+    url = f"{base}/completions"
+
+    # Prepare payload - use prompt_token_ids for completions endpoint
+    payload = {
+        "model": model or MODEL_NAME,
+        "prompt": prompt,
+        "max_tokens": max_tokens,
+        "temperature": temperature,
+        **kwargs,
+    }
+
+    # Make HTTP request
+    headers = {
+        "Authorization": f"Bearer {API_KEY}",
+        "Content-Type": "application/json",
+    }
+
+    last_exception = None
+    for attempt in range(max_retries + 1):
+        try:
+            with httpx.Client(timeout=timeout) as http_client:
+                response = http_client.post(url, json=payload, headers=headers)
+                response.raise_for_status()
+                result = response.json()
+
+            # Completions endpoint returns text directly
+            return result["choices"][0]["text"]
+        except (
+            httpx.HTTPStatusError,
+            httpx.RequestError,
+            httpx.TimeoutException,
+            httpx.NetworkError,
+        ) as e:
+            last_exception = e
+            if attempt < max_retries:
+                backoff = initial_backoff * (2**attempt)
+                time.sleep(backoff)
+                continue
+            raise
+
+    # This should never be reached, but type checker needs it
+    if last_exception:
+        raise last_exception
+    raise RuntimeError("Request failed unexpectedly")