Continual Learning Python Packages

honcho-ai

Memory library for building stateful agents

letta

Stateful agents that are like people, with memory, identity, and the ability to learn and adapt

stashai

Open-source infrastructure for agents that learn from experience. Capture production traces and turn lessons into reusable knowledge and skills.

capymoa

CapyMOA does efficient machine learning for data streams in Python. CapyMOA is a toolbox of methods and evaluators for: classification, regression, clustering, anomaly detection, semi-supervised learning, online continual learning, and drift detection for data streams.

honcho-cli

Memory library for building stateful agents

avalanche-lib

Avalanche: an End-to-End Library for Continual Learning based on PyTorch.

continuum

A clean and simple data loading library for Continual Learning

aberrant

Online Anomaly Detection Algorithms

reef-infra

Infrastructure for continually self‑improving agents

sillage

Your language model forgets everything. Sillage gives it a 4 MB memory that persists across sessions, no gradients, no fine-tuning, no growing index. Beats an unbounded kNN-LM at 1/13 the storage (perplexity 31 → 17). CPU-only, fully reproducible, four preprints with Zenodo DOIs.

feltstate

A character engine for AI agents — persistent affect, provenance-aware memory, gated proactivity, and an auditable forgetting lifecycle. Reference design.

darwin-memo

Self-curating memory for LLM agents: MeMo-style external memory kept honest by survival-based selection. No reward models, no judges.

arc-atlas

Atlas is infrastructure for continual learning in LLM agents.

moda-anthropic

Official Python SDK for Moda — the continual learning layer for AI agents.

moda-openai

Official Python SDK for Moda — the continual learning layer for AI agents.

moda-ai

Official Python SDK for Moda — the continual learning layer for AI agents.

biosnn-bus

脉冲神经网络认知原型：纯局部学习规则（无代理梯度）+ SNN 自主调度 LLM。Spiking neural network cognitive prototype — purely local learning rules (no surrogate gradients), SNN autonomously schedules an LLM, modality-plugin architecture. 研究原型，非生产框架 / research prototype, not a production framework.

adaptible

A small language model that keeps its conversations and retrains itself on its own corrections, on a laptop.

cordis-runtime

Helping AI agents remember what happened, learn from evidence, and approach the next task differently.

adaptergate

CI gate for per-tenant LoRA adapters that update online. Slice-level reject explain: tells you which behavioral cohort broke, not just that something did.

onad

Online Anomaly Detection Algorithms

cordis-memory

Helping AI agents remember what happened, learn from evidence, and approach the next task differently.

cordis-core

Helping AI agents remember what happened, learn from evidence, and approach the next task differently.

streamgen

Python framework for generating streams of labeled data.