llm-samplers 0.1.0

A library for advanced LLM sampling techniques

Samplers

A Python library for advanced LLM sampling techniques, providing a collection of sophisticated sampling methods for language models.

Features

• Temperature Scaling
• Top-K Sampling
• Top-P (Nucleus) Sampling
• Min-P Sampling
• Anti-Slop Sampling
• XTC (Exclude Top Choices) Sampling

Installation

From PyPI

pip install samplers

From Source

1. Clone the repository:

git clone https://github.com/iantimmis/samplers.git
cd samplers

2. Create and activate a virtual environment (recommended):

# Using venv
python -m venv .venv
source .venv/bin/activate  # On Unix/macOS
# or
.venv\Scripts\activate  # On Windows

# Using uv (recommended)
uv venv
source .venv/bin/activate  # On Unix/macOS
# or
.venv\Scripts\activate  # On Windows

3. Install the package in development mode:

# Using pip
pip install -e .

# Using uv (recommended)
uv pip install -e .

Development

Running Tests

The test suite uses pytest. To run the tests:

# Install test dependencies
pip install pytest

# Run all tests
python -m pytest tests/

# Run tests with verbose output
python -m pytest tests/ -v

# Run a specific test file
python -m pytest tests/test_temperature.py

Usage

from samplers import TemperatureSampler, TopKSampler, TopPSampler, MinPSampler, AntiSlopSampler, XTCSampler

# Initialize a sampler
sampler = TemperatureSampler(temperature=0.7)

# Use with a HuggingFace model
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("gpt2")
tokenizer = AutoTokenizer.from_pretrained("gpt2")

# Generate text with the sampler
input_text = "Once upon a time"
input_ids = tokenizer.encode(input_text, return_tensors="pt")
output_ids = sampler.sample(model, input_ids)
generated_text = tokenizer.decode(output_ids[0])

Available Samplers

Temperature Scaling

Adjusts the "sharpness" of the probability distribution:

• Low temperature (<1.0): More deterministic, picks high-probability tokens
• High temperature (>1.0): More random, flatter distribution

Top-K Sampling

Considers only the 'k' most probable tokens, filtering out unlikely ones.

Top-P (Nucleus) Sampling

Selects the smallest set of tokens whose cumulative probability exceeds threshold 'p'.

Min-P Sampling

Dynamically adjusts the sampling pool size based on the probability of the most likely token.

Anti-Slop

Down-weights probabilities at word & phrase level, using backtracking to retry with adjusted probabilities.

XTC (Exclude Top Choices)

Enhances creativity by nudging the model away from its most predictable choices.

License

MIT License