Metadata-Version: 2.4
Name: tinker_cookbook
Version: 0.2.1
Summary: Implementations of post-training algorithms using the Tinker API
Project-URL: Homepage, https://thinkingmachines.ai/tinker
Project-URL: Repository, https://github.com/thinking-machines-lab/tinker-cookbook
Project-URL: Documentation, https://tinker-docs.thinkingmachines.ai/
Author-email: Tinker authors <tinker@thinkingmachines.ai>
License-Expression: Apache-2.0
License-File: LICENSE
Requires-Python: >=3.11
Requires-Dist: aiohttp>=3.9.0
Requires-Dist: anyio>=4.0.0
Requires-Dist: blobfile>=3.0.0
Requires-Dist: chz>=0.4.0
Requires-Dist: cloudpickle>=3.0.0
Requires-Dist: datasets>=2.14.0
Requires-Dist: huggingface-hub>=0.20.0
Requires-Dist: numpy>=1.24.0
Requires-Dist: pillow>=10.0.0
Requires-Dist: pydantic>=2.0.0
Requires-Dist: rich>=13.0.0
Requires-Dist: safetensors>=0.4.0
Requires-Dist: termcolor>=2.0.0
Requires-Dist: tiktoken>=0.12.0
Requires-Dist: tinker>=0.9.0
Requires-Dist: torch>=2.0
Requires-Dist: tqdm>=4.60.0
Requires-Dist: transformers<=5.3.0,>=4.57.6
Provides-Extra: all
Requires-Dist: chromadb>=1.0.0; extra == 'all'
Requires-Dist: google-genai>=1.0.0; extra == 'all'
Requires-Dist: huggingface-hub>=0.20.0; extra == 'all'
Requires-Dist: inspect-ai>=0.3.100; extra == 'all'
Requires-Dist: inspect-evals>=0.3.106; extra == 'all'
Requires-Dist: litellm!=1.82.7,!=1.82.8,>=1.80.0; extra == 'all'
Requires-Dist: math-verify>=0.5.0; extra == 'all'
Requires-Dist: modal>=1.0.0; extra == 'all'
Requires-Dist: neptune-scale>=0.27.0; extra == 'all'
Requires-Dist: openai>=1.0.0; extra == 'all'
Requires-Dist: plotly>=5.0.0; extra == 'all'
Requires-Dist: pylatexenc>=2.0; extra == 'all'
Requires-Dist: sympy>=1.12.0; extra == 'all'
Requires-Dist: textarena>=0.7.4; extra == 'all'
Requires-Dist: trackio<1.0.0; extra == 'all'
Requires-Dist: verifiers<0.1.10,>=0.1.9; extra == 'all'
Requires-Dist: wandb>=0.16.0; extra == 'all'
Provides-Extra: dev
Requires-Dist: pyright>=1.1.300; extra == 'dev'
Requires-Dist: pytest-asyncio>=0.23.0; extra == 'dev'
Requires-Dist: pytest-timeout>=2.0.0; extra == 'dev'
Requires-Dist: pytest>=8.0.0; extra == 'dev'
Requires-Dist: ruff>=0.4.0; extra == 'dev'
Provides-Extra: inspect
Requires-Dist: inspect-ai>=0.3.100; extra == 'inspect'
Requires-Dist: inspect-evals>=0.3.106; extra == 'inspect'
Provides-Extra: litellm
Requires-Dist: litellm!=1.82.7,!=1.82.8,>=1.80.0; extra == 'litellm'
Provides-Extra: math-rl
Requires-Dist: math-verify>=0.5.0; extra == 'math-rl'
Requires-Dist: pylatexenc>=2.0; extra == 'math-rl'
Requires-Dist: sympy>=1.12.0; extra == 'math-rl'
Provides-Extra: modal
Requires-Dist: modal>=1.0.0; extra == 'modal'
Provides-Extra: multiplayer-rl
Requires-Dist: textarena>=0.7.4; extra == 'multiplayer-rl'
Provides-Extra: neptune-scale
Requires-Dist: neptune-scale>=0.27.0; extra == 'neptune-scale'
Provides-Extra: trackio
Requires-Dist: trackio<1.0.0; extra == 'trackio'
Provides-Extra: vector-search
Requires-Dist: chromadb>=1.0.0; extra == 'vector-search'
Requires-Dist: google-genai>=1.0.0; extra == 'vector-search'
Requires-Dist: huggingface-hub>=0.20.0; extra == 'vector-search'
Provides-Extra: verifiers
Requires-Dist: openai>=1.0.0; extra == 'verifiers'
Requires-Dist: verifiers<0.1.10,>=0.1.9; extra == 'verifiers'
Provides-Extra: wandb
Requires-Dist: plotly>=5.0.0; extra == 'wandb'
Requires-Dist: wandb>=0.16.0; extra == 'wandb'
Description-Content-Type: text/markdown

<h1 align="center">Tinker Cookbook</h1>
<div align="center">
  <img src="assets/tinker-cover.png" width="60%" />
</div>

<div align="center">

[![pytest](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/pytest.yaml/badge.svg)](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/pytest.yaml)
[![pyright](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/pyright.yaml/badge.svg)](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/pyright.yaml)
[![smoke-test-recipes](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/smoke-test-recipes.yaml/badge.svg)](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/smoke-test-recipes.yaml)
[![PyPI](https://img.shields.io/pypi/v/tinker-cookbook)](https://pypi.org/project/tinker-cookbook/)

</div>

We provide two libraries for the broader community to customize their language models: `tinker` and `tinker-cookbook`.

- `tinker` is a training SDK for researchers and developers to fine-tune language models. You send API requests to us and we handle the complexities of distributed training.
- `tinker-cookbook` includes realistic examples of fine-tuning language models. It builds on the Tinker API and provides common abstractions to fine-tune language models.

## Installation

1. Sign up for Tinker [here](https://auth.thinkingmachines.ai/sign-up).
2. Once you have access, create an API key from the [console](https://tinker-console.thinkingmachines.ai) and export it as environment variable `TINKER_API_KEY`.
3. Install `tinker-cookbook` (includes the `tinker` SDK as a dependency):
   ```bash
   # Latest stable release from PyPI
   uv pip install tinker-cookbook

   # Or install the nightly build
   uv pip install 'tinker-cookbook @ git+https://github.com/thinking-machines-lab/tinker-cookbook.git@nightly'
   ```

## Tinker

Refer to the [docs](https://tinker-docs.thinkingmachines.ai/training-sampling) to start from basics.
Here we introduce a few Tinker primitives - the basic components to fine-tune LLMs:

```python
import tinker
service_client = tinker.ServiceClient()
training_client = service_client.create_lora_training_client(
  base_model="meta-llama/Llama-3.2-1B", rank=32,
)
training_client.forward_backward(...)
training_client.optim_step(...)
training_client.save_state(...)
training_client.load_state(...)

sampling_client = training_client.save_weights_and_get_sampling_client()
sampling_client.sample(...)
```

See [tinker_cookbook/recipes/sl_loop.py](tinker_cookbook/recipes/sl_loop.py) and [tinker_cookbook/recipes/rl_loop.py](tinker_cookbook/recipes/rl_loop.py) for minimal examples of using these primitives to fine-tune LLMs.

To download the weights of any model:
```python
rest_client = service_client.create_rest_client()
future = rest_client.get_checkpoint_archive_url_from_tinker_path(sampling_client.model_path)
with open(f"model-checkpoint.tar.gz", "wb") as f:
    f.write(future.result())
```

### Tinker Cookbook

Besides these primitives, we also offer **Tinker Cookbook** (a.k.a. this repo), a library of a wide range of abstractions to help you customize training environments.
[`tinker_cookbook/recipes/sl_basic.py`](tinker_cookbook/recipes/sl_basic.py) and [`tinker_cookbook/recipes/rl_basic.py`](tinker_cookbook/recipes/rl_basic.py) contain minimal examples to configure supervised learning and reinforcement learning.

We also include a wide range of more sophisticated examples in the [`tinker_cookbook/recipes/`](tinker_cookbook/recipes/) folder:
1. **[Chat supervised learning](tinker_cookbook/recipes/chat_sl/)**: supervised fine-tuning on conversational datasets like Tulu3.
2. **[Math reasoning](tinker_cookbook/recipes/math_rl/)**: improve LLM reasoning capability by rewarding it for answering math questions correctly.
3. **[Preference learning](tinker_cookbook/recipes/preference/)**: showcase a three-stage RLHF pipeline: 1) supervised fine-tuning, 2) learning a reward model, 3) RL against the reward model.
4. **[Tool use](tinker_cookbook/recipes/search_tool/)**: train LLMs to better use retrieval tools to answer questions more accurately.
5. **[Prompt distillation](tinker_cookbook/recipes/prompt_distillation/)**: internalize long and complex instructions into LLMs.
6. **[Multi-Agent](tinker_cookbook/recipes/multiplayer_rl/)**: optimize LLMs to play against another LLM or themselves.

These examples are located in each subfolder, and their `README.md` files will walk you through the key implementation details, the commands to run them, and the expected performance.

### Documentation

The `docs/` directory contains a mirror of the Tinker documentation. These files are synced from our internal documentation site.

**Note:** The documentation files use MDX format (Markdown with JSX), which includes some syntax that isn't standard Markdown. You may see things like `import` statements, `<Callout>` components, or curly-brace expressions. These are artifacts of our documentation framework - the actual content should still be readable as Markdown.

If you find errors or want to improve the documentation, feel free to submit a PR editing files in `docs/`. We'll sync the changes back to our documentation site.

For the rendered documentation, visit [tinker-docs.thinkingmachines.ai](https://tinker-docs.thinkingmachines.ai).

### Import our utilities

Tinker cookbook includes several utilities. Here's a quick overview:
- [`renderers`](tinker_cookbook/renderers/) converts tokens from/to structured chat message objects
- [`hyperparam_utils`](tinker_cookbook/hyperparam_utils.py) helps calculate hyperparameters suitable for LoRAs
- [`evaluation`](tinker_cookbook/eval/evaluators.py) provides abstractions for evaluating Tinker models and [`inspect_evaluation`](tinker_cookbook/eval/inspect_evaluators.py) shows how to integrate with InspectAI to make evaluating on standard benchmarks easy.

## Claude Code Skills

Tinker Cookbook ships with [Claude Code skills](https://docs.anthropic.com/en/docs/claude-code/skills) that teach Claude how to use the Tinker API — SFT, RL, DPO, renderers, environments, and more. Install them globally so Claude can help you in any project:

```
/plugin marketplace add thinking-machines-lab/tinker-cookbook
```

Once installed, use `/tinker-sft`, `/tinker-grpo`, `/tinker-setup`, etc. in Claude Code. Skills update automatically from the repo.

## Development Setup

```bash
uv sync --extra dev
pre-commit install
```

This installs dev dependencies and registers pre-commit hooks that run `ruff` formatting and linting on every commit. CI enforces these checks on all pull requests.

## Contributing

This project is built in the spirit of open science and collaborative development. We believe that the best tools emerge through community involvement and shared learning.

We welcome PR contributions after our private beta is over. If you have any feedback, please email us at tinker@thinkingmachines.ai.

## Citation
If you use Tinker for your research, please cite it as:
```
Thinking Machines Lab, 2025. Tinker. https://thinkingmachines.ai/tinker/.
```

Or use this BibTeX citation:
```
@misc{tml2025tinker,
  author = {Thinking Machines Lab},
  title = {Tinker},
  year = {2025},
  url = {https://thinkingmachines.ai/tinker/},
}
```
