Thompson Sampling Python Packages

kalibr

Stop overpaying to run your agents. Kalibr routes every request to lower-cost model and tool paths without degrading performance.

pybandits

Python library for Multi-Armed Bandits

orts

Odds-Ratio Thompson Sampling: bandits and A/B tests whose baseline keeps moving. Paper: arXiv 2609.19709

oat-llm

🌾 OAT: A research-friendly framework for LLM online alignment, including reinforcement learning, preference learning, etc.

mabalgs

:bust_in_silhouette: Multi-Armed Bandit Algorithms Library (MAB) :cop:

hypotree

Memory that forgets: a persistent, self-revising belief-state MCP server and Python API for agentic R&D. Uses a hypothesis DAG to auto-prune dead ends and deduce truth without extra probes.

banditry

Python library for contextual bandits and Bayesian optimization over mixed design spaces, featuring GP-UCB, Thompson sampling, Gaussian processes, MCMC, and evolutionary acquisition optimization.

onn

Online Deep Learning: Learning Deep Neural Networks on the Fly / Non-linear Contextual Bandit Algorithm (ONN_THS)

orallexa-marketing-agent

AI marketing agent. Submit project once → auto-generate platform-native posts for 12 platforms (X, Reddit, HN, Dev.to, 小红书, ...). Bandit variant prediction. pip · MIT.

autoresearch-prompt-manager

Autonomous prompt management with versioning, A/B experiments, metrics, and LLM-driven optimization. Built on agno.

llm-cost-optimizer

Stop overpaying to run your agents. Kalibr routes every request to lower-cost model and tool paths without degrading performance.

llm-cost-reduction

Stop overpaying to run your agents. Kalibr routes every request to lower-cost model and tool paths without degrading performance.

marketing-agent-example

Autonomous prompt management with versioning, A/B experiments, metrics, and LLM-driven optimization. Built on agno.

autoresearcher-shonku

Autonomous prompt management with versioning, A/B experiments, metrics, and LLM-driven optimization. Built on agno.

shonku

Autonomous prompt management with versioning, A/B experiments, metrics, and LLM-driven optimization. Built on agno.

tsroots

TSRoots: A Python package for efficient Gaussian process Thompson sampling in Bayesian optimization via rootfinding.

retrieval-reputation-layer

Reputation layer for any retriever: turns verified downstream outcomes into a ranking signal (Beta counters, Thompson sampling, staleness decay). LangChain and LlamaIndex adapters.

dynamic-pricing-lab

Interactive in-browser simulator: dynamic pricing with demand learning, forward-looking customers, and advertising

fonz-dial

Online weight optimization via Thompson Sampling — learns optimal configurations from outcome feedback

kusp-dial

Online weight optimization via Thompson Sampling — learns optimal configurations from outcome feedback