Metadata-Version: 2.4
Name: bolt-bench
Version: 0.1.0
Summary: Benchmarks for BO on LLM Tasks
Author: Ruth Chew
License: MIT License
        
        Copyright (c) 2026 Ruth
        
        Permission is hereby granted, free of charge, to any person obtaining a copy
        of this software and associated documentation files (the "Software"), to deal
        in the Software without restriction, including without limitation the rights
        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
        copies of the Software, and to permit persons to whom the Software is
        furnished to do so, subject to the following conditions:
        
        The above copyright notice and this permission notice shall be included in all
        copies or substantial portions of the Software.
        
        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
        SOFTWARE.
License-File: LICENSE
Keywords: Bayesian optimization,Benchmark
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Requires-Python: >=3.10
Requires-Dist: botorch>=0.16.1
Requires-Dist: datasets
Requires-Dist: huggingface-hub>=0.36.0
Requires-Dist: numpy>=2.2.0
Requires-Dist: pandas
Requires-Dist: safetensors>=0.7.0
Requires-Dist: torch>=2.9.1
Provides-Extra: dev
Requires-Dist: pytest-cov; extra == 'dev'
Requires-Dist: pytest>=7.0; extra == 'dev'
Requires-Dist: ruff; extra == 'dev'
Provides-Extra: docs
Requires-Dist: mkdocs-material; extra == 'docs'
Requires-Dist: mkdocstrings[python]; extra == 'docs'
Requires-Dist: pygments; extra == 'docs'
Description-Content-Type: text/markdown

<img src="bolt_logo.png" width="300"/>

[![CI](https://github.com/chewwt/bolt/actions/workflows/ci.yml/badge.svg)](https://github.com/chewwt/bolt/actions/workflows/ci.yml)
[![codecov](https://codecov.io/gh/chewwt/bolt/branch/main/graph/badge.svg)](https://codecov.io/gh/chewwt/bolt)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
[![Python](https://img.shields.io/badge/python-3.10%20%7C%203.11%20%7C%203.12-blue.svg)](https://www.python.org/)

A benchmark suite for Bayesian optimization of expensive LLM tasks. Each problem is backed by a pretrained neural-network surrogate or tabular data from real LLM experiments, so evaluations are fast and reproducible without running real LLM training.


## Documentation

Full documentation is available at [bolt-bench.readthedocs.io](https://bolt-bench.readthedocs.io/en/latest/).

## Installation

`pip install bolt-bench`

## Quick Start

```python
import torch
from bolt import HPO

# 7-dim HPO problem: returns a scalar surrogate of eval score
prob = HPO(noise_std=0.001, negate=False)

X = torch.Tensor([[0, 2, 2, 2, 0.5, 30, 2]])  # one candidate configuration
y = prob(X)  # shape: (1,)
```


## Problems


| Problem | Class | Dims | Notes |
|---|---|---|---|
| HPO | `HPO` | 7 | mixed params (continuous, discrete, categorical) |
| HPO multi-fidelity (token) | `HPOMultiFidelityToken` | 8 | mixed params (continuous, discrete, categorical), fidelity: continuous ∈ [0, 1] (training tokens) |
| HPO multi-fidelity (model) | `HPOMultiFidelityModel` | 8 | mixed params (continuous, discrete, categorical), fidelity: discrete ∈ {0, 1} (model size) |
| Data mixture | `DMCurriculum` | 6 | two simplex constraints |
| Data mixture MO | `DMCurriculumMO` | 6 | two simplex constraints, multi-objective (3) |
| Data mixture with heteroscedastic noise | `DMCurriculumHet` | 6 | two simplex constraints, heteroscedastic noise |
| Prompt optimization (128-dim) | `PO128` | 128 | discrete candidate set |
| Prompt optimization (256-dim) | `PO256` | 256 | discrete candidate set |
| Prompt optimization (512-dim) | `PO512` | 512 | discrete candidate set |
| Prompt optimization (768-dim) | `PO768` | 768 | discrete candidate set |
